Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statement (IDS) documents submitted on June 16, 2023; November 28, 2023; August 29, 2024; and June 27, 2025 were filed and considered by the examiner. The submission is in compliance with the provisions of 37 CFR 1.97.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claim 1, recites the term " the artificial neural network" in the limitation “determine an inference distribution ratio of the artificial neural network of each of the plurality of devices…” without mentioning an artificial neural network model before this. There is insufficient antecedent basis for this limitation in claim 1. Dependent claims 3, 5, and 6 recite the same limitation and are rejected for the same reasons as hereinabove.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea (mental process or math concept) without significantly more.
Claim 1:
Regarding claim 1, in step 1 of the 101-analysis set forth in MPEP 2106, the claim recites
“an electronic device, comprising: a memory storing a state inference model, and at least one instruction; a transceiver; and at least one processor configured to execute the at least one instruction to: obtain, via the transceiver, first state information of each of a plurality of devices at a first time point, obtain second state information of each of the plurality of devices at a second time point that is a preset time interval after the first time point, by inputting the first state information to the state inference model, and determine an inference distribution ratio of the artificial neural network of each of the plurality of devices, based on the second state information of each of the plurality of devices, wherein the electronic device is determined from among the plurality of devices, based on network states of the plurality of devices,” and a device or machine is one of the four statutory categories of invention.
In step 2A prong 1 of the 101-analysis set forth in the MPEP 2106, the examiner has determined that the following limitations recite a process that, under the broadest reasonable interpretation, covers a mental process but for recitation of generic computer components:
and determine an inference distribution ratio of the artificial neural network of each of the plurality of devices, based on the second state information of each of the plurality of devices, (This is considered both a mental process and a mathematical concept, a person can mentally evaluate and determine an inference distribution ratio with calculations based on the second state information of each device , see MPEP 2106.04(a)(2)(III)), (This is also considered a mathematical relationship, mathematical formula or equation, or mathematical calculation, see in paragraphs [0087-0089] from the specification, describing “The inference ratio calculator 123 according to an embodiment of the disclosure may normalize an inverse number of an elapsed time of each device, as in Equation 1 below.
PNG
media_image1.png
8
9
media_image1.png
Greyscale
Also, the normalized inverse number of the elapsed time may be determined as an inference distribution ratio (r) of the artificial neural network”, see MPEP 2106.04(a)(2), subsection I),
If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process or as a mathematical concept, but for the recitation of generic computer components, then it falls within the mental process or mathematical concept groupings of abstract ideas. Accordingly, the claim “recites” an abstract idea.
In step 2A prong 2 of the 101-analysis set forth in MPEP 2106, the examiner has determined that the following additional elements do not integrate this judicial exception into a practical application:
An electronic device, comprising: a memory storing a state inference model, and at least one instruction; a transceiver; and at least one processor configured to execute the at least one instruction, (This is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
to: obtain, via the transceiver, first state information of each of a plurality of devices at a first time point, obtain second state information of each of the plurality of devices at a second time point that is a preset time interval after the first time point, by inputting the first state information to the state inference model, (In step 2A, prong 2, obtaining and inputting state information recites mere data gathering, which are considered insignificant extra-solution activities – see MPEP 2106.05(g)),
… wherein the electronic device is determined from among the plurality of devices, based on network states of the plurality of devices, (This is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f))
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is “directed” to an abstract idea.
In step 2B of the 101-analysis set forth in the 2019 PEG, the examiner has determined that the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
As discussed above, additional element ii and iv recite mere instructions to apply the judicial exception using generic computer components, which are not indicative of significantly more. The additional element iii recites mere data gathering, and is considered insignificant extra-solution activity. In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)),
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claim 2:
Regarding claim 2, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis applied to claim 1. Further, claim 2 recites the following additional element:
The electronic device of claim 1, wherein each of the first state information and the second state information comprises at least one of a usage rate of a central processing unit (CPU), a usage rate of a graphics processing unit (GPU), a temperature of the CPU, a temperature of the GPU, the number of executed applications, or an elapsed time of each of the plurality of devices, (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 3:
Regarding claim 3, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis applied to claim 1. Further, claim 3 recites the following abstract idea:
The electronic device of claim 1, wherein the second state information comprises an elapsed time, and the at least one processor is further configured to execute the at least one instruction to: normalize an inverse number of the elapsed time of each of the plurality of devices, (This is considered a mathematical relationship, mathematical formula or equation, or mathematical calculation, see in paragraphs [0087-0089] from the specification, describing “The inference ratio calculator 123 according to an embodiment of the disclosure may normalize an inverse number of an elapsed time of each device, as in Equation 1 below.
PNG
media_image1.png
8
9
media_image1.png
Greyscale
Also, the normalized inverse number of the elapsed time may be determined as an inference distribution ratio (r) of the artificial neural network” and recites a mathematical concept, see MPEP 2106.04(a)(2), subsection I),
and determine the normalized inverse number of the elapsed time, as the inference distribution ratio of the artificial neural network of each of the plurality of devices, (This is considered a mental process, since a person can mentally evaluate and determine a normalized inverse number of the elapsed time as an inference distribution ratio of the artificial neural network for each of the devices, see MPEP 2106.04(a)(2)(III)), (This is also considered a mathematical relationship, mathematical formula or equation, or mathematical calculation, see in paragraphs [0087-0089] from the specification, describing “The inference ratio calculator 123 according to an embodiment of the disclosure may normalize an inverse number of an elapsed time of each device, as in Equation 1 below.
PNG
media_image1.png
8
9
media_image1.png
Greyscale
Also, the normalized inverse number of the elapsed time may be determined as an inference distribution ratio (r) of the artificial neural network” and recites a mathematical concept, see MPEP 2106.04(a)(2), subsection I),
If the claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process or as a mathematical concept, but for the recitation of generic computer components, then it falls within the mental process or mathematical concept groupings of abstract ideas. Accordingly, the claim “recites” an abstract idea.
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 4:
Regarding claim 4, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis applied to claim 1. Further, claim 4 recites the following additional elements:
The electronic device of claim 1, wherein the at least one processor is further configured to execute the at least one instruction to: obtain third state information, (In step 2A, prong2, obtaining third state information recites mere data gathering, which is considered insignificant extra-solution activity, from the specification in paragraph [0093] which notes “referring to FIG. 4, the state inference unit 121 may receive an input of first state information 450 at a first time point T with respect to the first device 200 of FIG. 2,” and later in paragraphs [0097-0098] states “FIG. 5 illustrates an example of an operation in which an electronic device infers second state information by receiving an input of first state information and third state information, according to an embodiment. Referring to FIG. 5, the state inference unit 121 may additionally receive an input of third state information 501 as well as the pieces of input state information 410, 430, and 450 of FIG. 4. The third state information 501 that is input at a first time point T may include whether a particular application App 1 is executed and whether a screen is turned on,” – see MPEP 2106.05(g),). In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)),
comprising at least one of whether a preset application is executed, whether a screen is turned on, or whether a camera is executed, at the first time point; (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
..and obtain the second state information based on additionally inputting the third state information to the state inference model, (In step 2A, prong2, obtaining third state information recites mere data gathering, which is considered insignificant extra-solution activity, from the specification in paragraph [0087] which notes “the inference ratio calculator 123 may receive an input of second state information of each device from the state inference unit 121,” – see MPEP 2106.05(g),). In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 5:
Regarding claim 5, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis applied to claim 1. Further, claim 5 recites the following additional element:
The electronic device of claim 1, wherein the at least one processor is further configured to execute the at least one instruction to transmit, via the transceiver, the determined inference distribution ratio and an inference start point of the artificial neural network to each of the plurality of devices, (In step 2A, prong2, transmitting information recites mere data gathering, which is considered insignificant extra-solution activity – see MPEP 2106.05(g),). In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 6:
Regarding claim 6, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis applied to claim 1. Further, claim 6 recites the following additional elements:
The electronic device of claim 1, wherein the at least one processor is further configured to execute the at least one instruction to: partition the artificial neural network according to the determined inference distribution ratio, (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
… and transmit, via the transceiver, the partitioned artificial neural network to each of the plurality of devices corresponding to the determined inference distribution ratio, (In step 2A, prong2, transmitting information from the partitioned artificial neural network recites mere data gathering, which is considered insignificant extra-solution activity – see MPEP 2106.05(g),). In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)),
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 7:
Regarding claim 7, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis applied to claim 1. Further, claim 7 recites the following additional element:
The electronic device of claim 1, wherein the state inference model is regression-trained based on an input of state information for training at a third time point and target state information at a fourth time point after a preset time interval from the third time point, (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 8:
Regarding claim 8, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis applied to claim 1. Further, claim 8 recites the following additional element:
The electronic device of claim 1, wherein the network states are network input/output (I/O) packet amounts of the plurality of devices based on test information received by a first device from among the plurality of devices excluding the first device, the first device being randomly selected from the plurality of devices, (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 9:
Regarding claim 9, it is dependent upon claim 8, and thereby incorporates the limitations of, and corresponding analysis applied to claim 8. Further, claim 9 recites the following additional element:
The electronic device of claim 8, wherein the electronic device is a candidate device connected to a wired network from among at least one candidate device that is selected from among the plurality of devices and has a network I/O packet amount equal to or smaller than a preset packet amount, (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 10:
Regarding claim 10, it is dependent upon claim 9, and thereby incorporates the limitations of, and corresponding analysis applied to claim 9. Further, claim 10 recites the following additional element:
The electronic device of claim 9, wherein the electronic device is a candidate device having a highest GPU throughput from among the at least one candidate device, (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 11:
Regarding claim 11, in step 1 of the 101-analysis set forth in MPEP 2106, the claim recites “a method, performed by an electronic device, comprising: obtaining first state information at a first time point from each of a plurality of devices; obtaining second state information of each of the plurality of devices at a second time point that is a preset time interval after the first time point, by inputting the first state information to a state inference model; and determining an inference distribution ratio of an artificial neural network of each of the plurality of devices, based on the second state information of each of the plurality of devices, wherein the electronic device is determined among the plurality of devices, based on network states of the plurality of devices”, and a method is one of the four statutory categories of invention.
In step 2A prong 1 of the 101-analysis set forth in the MPEP 2106, the examiner has determined that the following limitations recite a process that, under the broadest reasonable interpretation, covers a mental process or mathematical process but for recitation of generic computer components:
and determining an inference distribution ratio of an artificial neural network of each of the plurality of devices, based on the second state information of each of the plurality of devices, (This is considered both a mental process and a mathematical concept, a person can mentally evaluate and determine an inference distribution ratio with calculations based on the second state information of each device, see MPEP 2106.04(a)(2)(III)), (This is also considered a mathematical relationship, mathematical formula or equation, or mathematical calculation, see in paragraphs [0087-0089] from the specification, describing “The inference ratio calculator 123 according to an embodiment of the disclosure may normalize an inverse number of an elapsed time of each device, as in Equation 1 below.
PNG
media_image1.png
8
9
media_image1.png
Greyscale
Also, the normalized inverse number of the elapsed time may be determined as an inference distribution ratio (r) of the artificial neural network”, which recites a mathematical concept, see MPEP 2106.04(a)(2), subsection I),
If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process or as a mathematical concept, but for the recitation of generic computer components, then it falls within the mental process or mathematical concept groupings of abstract ideas. Accordingly, the claim “recites” an abstract idea.
In step 2A prong 2 of the 101-analysis set forth in MPEP 2106, the examiner has determined that the following additional elements do not integrate this judicial exception into a practical application:
a method, performed by an electronic device, comprising: obtaining first state information at a first time point from each of a plurality of devices; (In step 2A, prong 2, obtaining state information recites mere data gathering, which are considered insignificant extra-solution activities – see MPEP 2106.05(g)),
obtaining second state information of each of the plurality of devices at a second time point that is a preset time interval after the first time point, by inputting the first state information to a state inference model; (In step 2A, prong 2, obtaining second state information that is a preset time interval after the first time point, and inputting state information recites mere data gathering, which are considered insignificant extra-solution activities – see MPEP 2106.05(g)),
… wherein the electronic device is determined among the plurality of devices, based on network states of the plurality of devices, (This is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f))
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is “directed” to an abstract idea.
In step 2B of the 101-analysis set forth in the 2019 PEG, the examiner has determined that the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
As discussed above, additional element iv recites mere instructions to apply the judicial exception using generic computer components, which is not indicative of significantly more. The additional elements ii and iii recite mere data gathering, and are considered insignificant extra-solution activities. In step 2B, these insignificant extra-solution activities are well understood routine and conventional activities which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)),
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claim 12:
Regarding claim 12, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis applied to claim 11. Further, claim 12 recites the following additional element:
The method of claim 11, wherein each of the first state information and the second state information comprises at least one of a usage rate of a central processing unit (CPU), a usage rate of a graphics processing unit (GPU), a temperature of the CPU, a temperature of the GPU, the number of executed applications, or an elapsed time of each of the plurality of devices, (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 13:
Regarding claim 13, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis applied to claim 11. Further, claim 13 recites the following abstract idea:
The method of claim 11, wherein the second state information comprises an elapsed time, and the determining of the inference distribution ratio comprises: normalizing an inverse number of the elapsed time of each of the plurality of devices; (This is considered a mathematical relationship, mathematical formula or equation, or mathematical calculation, see in paragraphs [0087-0089] from the specification, describing “The inference ratio calculator 123 according to an embodiment of the disclosure may normalize an inverse number of an elapsed time of each device, as in Equation 1 below.
PNG
media_image1.png
8
9
media_image1.png
Greyscale
Also, the normalized inverse number of the elapsed time may be determined as an inference distribution ratio (r) of the artificial neural network”, see MPEP 2106.04(a)(2), subsection I),
and determining the normalized inverse number of the elapsed time, as the inference distribution ratio of the artificial neural network of each of the plurality of devices, (This is considered a mental process, since a person can mentally evaluate and determine a normalized inverse number of the elapsed time as an inference distribution ratio of the artificial neural network for each of the devices, see MPEP 2106.04(a)(2)(III)), (This is also considered a mathematical relationship, mathematical formula or equation, or mathematical calculation, see in paragraphs [0087-0089] from the specification, describing “The inference ratio calculator 123 according to an embodiment of the disclosure may normalize an inverse number of an elapsed time of each device, as in Equation 1 below.
PNG
media_image1.png
8
9
media_image1.png
Greyscale
Also, the normalized inverse number of the elapsed time may be determined as an inference distribution ratio (r) of the artificial neural network”, see MPEP 2106.04(a)(2), subsection I),
If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process or as a mathematical concept, but for the recitation of generic computer components, then it falls within the mental process or mathematical concept groupings of abstract ideas. Accordingly, the claim “recites” an abstract idea.
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 14:
Regarding claim 14, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis applied to claim 11. Further, claim 14 recites the following additional elements:
The method of claim 11, wherein the obtaining of the second state information comprises: obtaining third state information, (In step 2A, prong2, obtaining third state information recites mere data gathering, which is considered insignificant extra-solution activity, from the specification in paragraph [0093] which notes “referring to FIG. 4, the state inference unit 121 may receive an input of first state information 450 at a first time point T with respect to the first device 200 of FIG. 2,” and later in paragraphs [0097-0098] states “FIG. 5 illustrates an example of an operation in which an electronic device infers second state information by receiving an input of first state information and third state information, according to an embodiment. Referring to FIG. 5, the state inference unit 121 may additionally receive an input of third state information 501 as well as the pieces of input state information 410, 430, and 450 of FIG. 4. The third state information 501 that is input at a first time point T may include whether a particular application App 1 is executed and whether a screen is turned on,” – see MPEP 2106.05(g),). In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)).
comprising at least one of whether a preset application is executed, whether a screen is turned on, or whether a camera is executed, at the first time point; (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
.. and obtaining the second state information based on additionally inputting the third state information to the state inference model, (In step 2A, prong2, obtaining third state information recites mere data gathering, which is considered insignificant extra-solution activity, from the specification in paragraph [0087] which notes “the inference ratio calculator 123 may receive an input of second state information of each device from the state inference unit 121,” – see MPEP 2106.05(g),). In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 15:
Regarding claim 15, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis applied to claim 11. Further, claim 15 recites the following additional element:
The method of claim 11, further comprising: transmitting the determined inference distribution ratio and an inference start point of the artificial neural network to each of the plurality of devices, (In step 2A, prong2, transmitting information recites mere data gathering, which is considered insignificant extra-solution activity – see MPEP 2106.05(g),). In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)),
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 16:
Regarding claim 16, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis applied to claim 11. Further, claim 16 recites the following additional element:
The method of claim 11, further comprising: partitioning the artificial neural network according to the determined inference distribution ratio; and transmitting the partitioned artificial neural network to each of the plurality of devices corresponding to the determined inference distribution ratio, (In step 2A, prong2, transmitting information recites mere data gathering, which is considered insignificant extra-solution activity – see MPEP 2106.05(g),). In step 2B, this insignificant extra-solution activity is well understood routine and conventional activity which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)),
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 17:
Regarding claim 17, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis applied to claim 11. Further, claim 17 recites the following additional element:
The method of claim 11, wherein the state inference model is regression- trained based on an input of state information for training at a third time point and target state information at a fourth time point after a preset time interval from the third time point, (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 18:
Regarding claim 18, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis applied to claim 11. Further, claim 18 recites the following additional element:
The method of claim 11, wherein the network states are network input/output (I/O) packet amounts of the plurality of devices based on test information received by a first device from among the plurality of devices excluding the first device, the first device being randomly selected from the plurality of devices, (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 19:
Regarding claim 19, it is dependent upon claim 18, and thereby incorporates the limitations of, and corresponding analysis applied to claim 18. Further, claim 19 recites the following additional element:
The method of claim 18, wherein the electronic device is a candidate device connected to a wired network from among at least one candidate device that is selected from among the plurality of devices and has a network I/O packet amount equal to or smaller than a preset packet amount, (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)), (In step 2B, this is also considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 20:
Regarding claim 20, in step 1 of the 101-analysis set forth in MPEP 2106, the claim recites “a non-transitory computer readable medium for storing computer readable program code or instructions which are executable by a processor to perform a method, the method comprising: obtaining first state information at a first time point from each of devices comprising the electronic device; obtaining second state information of each of the devices at a second time point that is a preset time interval after the first time point, by inputting the first state information to a state inference model; and determining an inference distribution ratio of an artificial neural network of each of the plurality of devices, based on the second information of each of the plurality of devices, wherein the electronic device is determined from among the plurality of devices, based on network states of the plurality of devices,” and a non-transitory computer readable medium for storing computer readable program code or instructions, is considered a machine and is one of the four statutory categories of invention.
In step 2A prong 1 of the 101-analysis set forth in the MPEP 2106, the examiner has determined that the following limitations recite a process that, under the broadest reasonable interpretation, covers a mental process but for recitation of generic computer components:
and determining an inference distribution ratio of an artificial neural network of each of the plurality of devices, based on the second information of each of the plurality of devices, (This is considered both a mental process and a mathematical concept, a person can mentally evaluate and determine an inference distribution ratio with calculations based on the second state information of each device , see MPEP 2106.04(a)(2)(III)), (This is also considered a mathematical relationship, mathematical formula or equation, or mathematical calculation, see in paragraphs [0087-0089] from the specification, describing “The inference ratio calculator 123 according to an embodiment of the disclosure may normalize an inverse number of an elapsed time of each device, as in Equation 1 below.
PNG
media_image1.png
8
9
media_image1.png
Greyscale
Also, the normalized inverse number of the elapsed time may be determined as an inference distribution ratio (r) of the artificial neural network”, which recites a mathematical concept, see in MPEP 2106.04(a)(2), subsection I),
If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process or as a mathematical concept, but for the recitation of generic computer components, then it falls within the mental process or mathematical concept groupings of abstract ideas. Accordingly, the claim “recites” an abstract idea.
In step 2A prong 2 of the 101-analysis set forth in MPEP 2106, the examiner has determined that the following additional elements do not integrate this judicial exception into a practical application:
A non-transitory computer readable medium for storing computer readable program code or instructions which are executable by a processor to perform a method, (This is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
the method comprising: obtaining first state information at a first time point from each of devices comprising the electronic device; (In step 2A, prong 2, obtaining state information recites mere data gathering, which are considered insignificant extra-solution activities – see MPEP 2106.05(g)),
obtaining second state information of each of the devices at a second time point that is a preset time interval after the first time point, by inputting the first state information to a state inference model; (In step 2A, prong 2, obtaining and inputting state information recites mere data gathering, which are considered insignificant extra-solution activities – see MPEP 2106.05(g)),
wherein the electronic device is determined from among the plurality of devices, based on network states of the plurality of devices, (This is considered mere instructions to apply an exception using generic computer – see MPEP 2106.05(f)),
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is “directed” to an abstract idea.
In step 2B of the 101-analysis set forth in the 2019 PEG, the examiner has determined that the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
As discussed above, additional element ii and v recite mere instructions to apply the judicial exception using generic computer components, which are not indicative of significantly more. The additional elements iii and iv recite mere data gathering, and are considered insignificant extra-solution activities. In step 2B, these insignificant extra-solution activities are well understood routine and conventional activities which includes receiving or transmitting data over a network from court case Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607, 610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016), – see MPEP 2106.05(d) (II)(i)),
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1, 5, 6, 11, 15, 16, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Kim J. et al. (Pub. No. KR20180071932 A), published on June 28, 2018, (hereafter, KIM J.), in view of Jaiswal A. et al. in “Accurate Device Temperature Forecasting using Recurrent Neural Network for Smartphone Thermal Management,” published on September 20, 2021, available on https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9533732, (hereafter, JAISWAL), further in view of Mao, J. et al. in "MoDNN: Local distributed mobile computing system for Deep Neural Network," published March 27-31, 2017 for a conference, made available online on May 15, 2017, available at https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7927211&tag=1, (hereafter, MAO).
Claim 1:
Regarding claim 1, KIM J. teaches “an electronic device, comprising: a memory storing a state inference model, and at least one instruction;”
See KIM J. in paragraph [0188] describe “further, when the data recognition model is learned, the model learning unit 1010-4 can store the learned data recognition model. In this case, the model learning unit 1010-4 can store the learned data recognition model in the memory of the server 130.” Here, KIM J. describes the model (which refers to a state inference model) is stored in a memory. See paragraph [0025] from KIM J. for more information.
Further, see KIM J. in paragraph [0061] describe “the storage unit 120 is configured to store various programs and information required for the control of the server 100. For example, the storage unit 120 may be implemented as a nonvolatile memory, a volatile memory, a flash memory, a hard disk drive (HDD), or a solid state drive (SSD). Meanwhile, the storage unit may be implemented not only as a storage medium in the server 100, but also as an external storage medium, such as a micro SD card, a USB memory, or an external server.” Here, KIM J. teaches that the storage unit contains a memory that stores various programs and information required for the control of the server 100, where information can include at least one instruction. See KIM J. in paragraphs [0173] and [0198] for more information.
Further, KIM J. teaches “a transceiver;”
See KIM J. in paragraph [0140] describe “on the other hand, the server 100 can transmit not only the information about the error device but also the information about the determined operation status to the user terminal device. 4, the server 100 receives information indicating that an operation state in which cooling of the air conditioner 41 has been activated has occurred, and transmits information indicating that an error has occurred to the first temperature sensor 42, To the terminal device.” Here, the examiner construes transceiver to have functions both as a receiver and transmitter to carry and deliver information. Here, KIM J. mentions the server 100 has both receiver and transmitter roles which are included in the functions of a transceiver. See KIM J. in paragraph [0150] for more information.
Further, KIM J. teaches “and at least one processor configured to execute the at least one instruction,”
See KIM J. in paragraph [0165] discuss “meanwhile, the processor 130 of the server according to the embodiment of the present disclosure can determine whether there is an error in a plurality of sensors and electronic devices by using a learning algorithm, and provide information on a determination result to a user.” Here, KIM J. teaches at least one processor configured to execute at least one instruction.
Further, KIM J. teaches “to: obtain, via the transceiver, first state information of each of a plurality of devices at a first time point,”
See KIM J. in paragraph [0019] describe “when it is determined that the operation status of the at least one electronic device is a new operation status, the processor may cause the sensing data received from each of the plurality of sensors to correspond to the new operation status at a time corresponding to the new operation status The reference sensing data may be stored in the storage unit.” Further, see KIM J. describe in paragraph [0025] “according to another aspect of the present invention, there is provided a method of controlling a server for managing a home network, the method comprising: storing operation conditions of at least one electronic device in the home network and reference sensor data for each sensor; Receiving the sensed data from the at least one electronic device,” Since a device can have at least a sensor, each of the plurality of sensors also relate to each of the devices. See paragraph [0047] in KIM J. for more information.
Further, see KIM J. mention in paragraph [0071] “depending on the client device, the status information may be transmitted to the server 100 continuously or at predetermined intervals. Alternatively, the processor 130 may request the status information to be transmitted to the client device through the communication unit 110, and the client device may transmit status information to the server 100 according to the request. Alternatively, the processor 130 may control the communication unit 110 to transmit a control command for performing a specific operation to the client apparatus, and the client apparatus may perform an operation on the received control command, State information can be transmitted as information.” Here, KIM J. shows status information is viewed as similar meaning as state information. Further, see KIM J. in paragraph [0092] mention “referring to FIG. 4, the cooling of the air conditioner 41 is activated, and the air conditioner 41 transmits status information to the server 100 to inform that the cooling has been performed. The status information may also include information on the cooling temperature. The processor 130 may determine an operation state that the cooling of the air conditioner 41 is activated based on the state information received from the air conditioner 41. The processor 130 may determine the operation state And analyzes the sensing data received from the sensor 44 and the temperature sensors 42 and 43. In particular, the processor 130 may calculate the time taken for the temperature sensor 42, 43 to detect a change in temperature from the cooling start time.” Here, KIM J. shows that the server 100 which acts as a transceiver, and obtains temperature measurements (i.e. first state information) at a start time (i.e. first time point).
However, KIM J. did not teach:
“obtain second state information of each of the plurality of devices at a second time point that is a preset time interval after the first time point, by inputting the first state information to the state inference model,”
“and determine an inference distribution ratio of the artificial neural network of each of the plurality of devices, based on the second state information of each of the plurality of devices, wherein the electronic device is determined from among the plurality of devices, based on network states of the plurality of devices.”
In an analogous system, JAISWAL teaches “obtain second state information of each of the plurality of devices at a second time point that is a preset time interval after the first time point, by inputting the first state information to the state inference model,”
See JAISWAL in page 5, section V. Proposed Method, paragraph 3, describe "The model needs to know the input shape. For this reason, the first layer in a Sequential model needs to receive information about it. Input needs to be in the form [samples, time steps, features], where 'samples' is the total number of data points, 'time steps' is the number of time-dependent steps, 'features' refers to the number of variables." Here, JAISWAL shows the inputting of a time step (i.e. time point). See JAISWAL in page 6, V. Proposed Method, column 1, paragraph 3, describe "Rectified Linear Activation Function (ReLU) is used to build the TFE model as it allows it to learn nonlinear dependencies and complex mapping functions ... Where,
Xt = input at current timestamp t
Yt = output at current timestamp t
Ht = output of LSTM block
Steps = 40 ", and refer to figures 9 and 10, and table IX for more details.
PNG
media_image2.png
240
704
media_image2.png
Greyscale
PNG
media_image3.png
892
1298
media_image3.png
Greyscale
Further, see JAISWAL describe in page 7, section VI. On-Device Results, first paragraph of “Apart from the flagship Samsung Galaxy S20, we have validated the TFE model across different Samsung smartphone devices having different form factor and internal designs, Samsung A50, A70 and Samsung Tablet Active 2.” Here, the time step mentioned by JAISWAL is interpreted by examiner to mean the time point. Here, JAISWAL shows that by inputting the first or initial elapsed time or first state information of Xt at current timestamp t, the temperature measurement (i.e. second state information) is obtained for each device (see table IX, where scenarios represent to each type of operation for various Samsung smartphone devices described in page 7, section VI ) at a second time point that is a preset time interval (from figure 9 shows set intervals of 30 seconds, 1 minute or 3 minutes) after the first time point.
Further, see JAISWAL in page 2, section II. Related Work, in second column, paragraphs 5-6, describe "certain limited exothermic components in the device have been utilized [10] to define the complex relationship between the various components inside of a mobile device and its surface temperature. Although comprehensive measurement results for a specific time window for predictions or for high load use cases like FastDischarge [15] and heavy high-end multiplayer games - PlayerUnknown's Battlegrounds (PUBG) [16] Counter - Strike: Global Offensive (CS GO) [17] may not have been calculated. Our prediction model is different from the above mentioned studies in that we use a complex Recurrent Neural Network - Long Short Term Memory, .... Also, ours is a multi-step model, which can be modified to accurately predict temperatures at multiple intervals at a time. While accurate predictions 5 seconds into the future have been made [13], the range of our prediction can be as long as 3 minutes, giving significant leverage to activate mitigation mechanisms well in advance." Here, JAISWAL describes predicting temperature, which is a state information, at multiple intervals at a time such as 5 seconds or 3 minutes into the future, and shows a specific time window or preset time interval after an initial or first time point.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the reference of KIM J. along with the teachings of JAISWAL by using the teachings of KIM J. of a device with a processor and memory that uses artificial neural network models, with JAISWAL’s teaching of using a device that obtains state information at time points.
One of ordinary skill in the art would be motivated to do so because by integrating JAISWAL’s framework into the system of KIM J., one with ordinary skill in the art would achieve the goal of providing a model that is “capable of providing accurate advance prediction of device surface temperature with average error rate of 0.64°C, which can efficiently handle temperature implications thereby enabling longer sustained device performance under 5G, gaming, multimedia and other heavy use cases. Clients can effectively use this thermal prediction to select the best available mitigation strategy,” (JAISWAL, page 7, section VII. Conclusion And Future Work).
However, KIM J. in view of JAISWAL did not explicitly teach “and determine an inference distribution ratio of the artificial neural network of each of the plurality of devices, based on the second state information of each of the plurality of devices, wherein the electronic device is determined from among the plurality of devices, based on network states of the plurality of devices,”
In an analogous art, MAO teaches “and determine an inference distribution ratio of the artificial neural network of each of the plurality of devices, based on the second state information of each of the plurality of devices, wherein the electronic device is determined from among the plurality of devices, based on network states of the plurality of devices,”
See MAO in abstract, page 1396 describes the model "MoDNN can partition already trained DNN models onto several mobile devices to accelerate DNN computations by alleviating device-level computing cost and memory usage." Here, MAO describes a model using deep neural networks or DNN (i.e. an artificial neural network) that partitions those models across multiple mobile devices (i.e. wherein the electronic device is determined from among the plurality of devices). In addition, MAO describes in page 1397 in section III. System Framework of MoDNN, first paragraph, and page 1400, section IV. Experiments, first paragraph for more information.
Later, see MAO in page 1397 section III, part A. Definition of Terminologies and Variables, MAO describes “Data Delivery Time: Data delivery time denotes the total time consumption for the data being transmitted between nodes. Data delivery time includes two parts: wakeup time and transmission time. Wakeup time represents the amount of time for the head of the data traveling from the sender to the receiver and transmission time denotes the amount of time for the receiver receiving from the first bit to the last bit of the data.” Here in page 1397, MAO is measuring two types of information for a first state, the first information is the wakeup time which is the elapsed time for data to travel from sender to the receiver, and a second information is the transmission time which is an elapsed time for receiver receiving data for each device. Both of these are considered to be a first state of a device. From the application’s specification in paragraph [0007], which states “each of the first state information and the second state information may include at least one of a usage rate of a central processing unit (CPU), a usage rate of a graphics processing unit (GPU), a temperature of the CPU, a temperature of the GPU, the number of executed applications, or an elapsed time of each of the plurality of devices”. Elapsed time is construed to mean any time that occurs regarding an operation of a device.
Further, see MAO in page 1399, section 3) Fine Grain Cross Partition (FGCP), paragraph 2, describe “In addition, FGCP needs to consider the discrepancy of execution time between the GO and the worker nodes during the offloading process, especially the data delivery time on the network between the worker node x and the GO, which can be conceptually formulated by:
PNG
media_image4.png
72
576
media_image4.png
Greyscale
where TPT is the mobile network throughput;” The examiner construes inference distribution ratio as any mathematical operation that distributes computation or resource usage among various devices. Here, MAO describes a first state, which is the delivery time incorporated into equation 2 (i.e. inference distribution ratio of the artificial neural network) from all devices with the numerator, worker node refers to a device, and TPT, a second state, measures a throughput of mobile networks in page 1399, and equation 2 can be applied across multiple devices using their respective neural network models described in abstract in page 1396. See algorithm 2 in page 1399 for more details.
Later, MAO describes in page 1400, section IV. Experiments, part A. Environment Setup and Testbench Selection that “The measured average WLAN wakeup time and transmission throughput are 54.7ms and 43.8Mbps, respectively.” TPT, which is a throughput measurement of mobile network, is interpreted as a measure of a device’s computation usage. Since each device is connected to a network, each device has their own throughput and computation usage (measured in page 1400 section IV in units Mbps), which is based on a second state.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of KIM J. and JAISWAL along with the teachings of MAO by using the teachings of KIM J. and JAISWAL of devices that obtain state information at time points, with MAO’s teaching of using an inference distribution ratio with state information on electronic devices.
One of ordinary skill in the art would be motivated to do so because by integrating MAO’s framework into the methods of KIM J. and JAISWAL, one with ordinary skill in the art would achieve the goal of having a model that “can accelerate the DNN computation by 2.17-4.28×. Besides the parallel execution, the performance speedup also partially comes from the reduction of the data delivery time,” (MAO, page 1396, abstract).
Claim 5:
Regarding claim 5, KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 1.
Further, KIM J. teaches “the electronic device of claim 1, wherein the at least one processor is further configured to execute the at least one instruction to transmit, via the transceiver,”
See KIM J. in paragraph [0140] describe “on the other hand, the server 100 can transmit not only the information about the error device but also the information about the determined operation status to the user terminal device. 4, the server 100 receives information indicating that an operation state in which cooling of the air conditioner 41 has been activated has occurred, and transmits information indicating that an error has occurred to the first temperature sensor 42, To the terminal device.” Here, the examiner construes transceiver to have functions both as a receiver and transmitter to carry and deliver information. Here, KIM J. mentions the server 100 has both receiver and transmitter roles which are included in the functions of a transceiver. See KIM J. in paragraph [0150] for more information.
Further, see KIM J. in paragraph [0165] discuss “meanwhile, the processor 130 of the server according to the embodiment of the present disclosure can determine whether there is an error in a plurality of sensors and electronic devices by using a learning algorithm, and provide information on a determination result to a user.” See KIM J. in paragraph [0058] mention “referring to FIG. 2, the server 100 includes a communication unit 110, a storage unit 120, and a processor 130.” KIM J. connects a processor with a transceiver (server 100). Here, KIM J. teaches at least one processor configured to execute at least one instruction. See KIM J. in paragraphs [0021-0024] for more information.
Further, KIM J. in view of JAISWAL, further in view of MAO, teaches “the determined inference distribution ratio and an inference start point of the artificial neural network to each of the plurality of devices,”
See MAO in abstract, page 1396 describes the model "MoDNN can partition already trained DNN models onto several mobile devices to accelerate DNN computations by alleviating device-level computing cost and memory usage." Here, MAO describes a model using deep neural networks or DNN (i.e. an artificial neural network) that partitions those models across multiple mobile devices (i.e. wherein the electronic device is determined from among the plurality of devices). In addition, MAO describes in page 1397 in section III. System Framework of MoDNN, first paragraph, and page 1400, section IV. Experiments, first paragraph for more information.
Later, see MAO in page 1397 section III, part A. Definition of Terminologies and Variables, MAO describes “Data Delivery Time: Data delivery time denotes the total time consumption for the data being transmitted between nodes. Data delivery time includes two parts: wakeup time and transmission time. Wakeup time represents the amount of time for the head of the data traveling from the sender to the receiver and transmission time denotes the amount of time for the receiver receiving from the first bit to the last bit of the data.” In page 1397, MAO is measuring a wakeup time which is the elapsed time for data to travel from sender to the receiver, and a transmission time which is an elapsed time for receiver receiving data for each device. Both of these are considered to be a first state of a device. MAO also mentions that the transmission time shows the amount of time for a receiver device to receive the first bit to last bit of data, with first being an initial time (i.e. an inference start point) to gather data, and is considered to be a starting time point for inference using the MoDNN neural network model.
Further, see MAO in table I. Overall Evaluation Of MoDNN With 2-4 Worker Nodes in page 1401, showing a start time (i.e. an inference start point) for the model to run inference tasks:
PNG
media_image5.png
226
588
media_image5.png
Greyscale
Further, see MAO in page 1399, section 3) Fine Grain Cross Partition (FGCP), paragraph 2, describe “In addition, FGCP needs to consider the discrepancy of execution time between the GO and the worker nodes during the offloading process, especially the data delivery time on the network between the worker node x and the GO, which can be conceptually formulated by:
PNG
media_image4.png
72
576
media_image4.png
Greyscale
where TPT is the mobile network throughput;” The examiner construes inference distribution ratio as any mathematical operation that distributes computation or resource usage among various devices. Here, MAO describes that determining an estimated time or variable Initial_ET for that device to implement a workload from a first state, which is the elapsed delivery time of all devices incorporated into equation 2 (i.e. inference distribution ratio) in the numerator calculated from algorithm 2, and another state, TPT, which measures a computation usage throughput of mobile networks in page 1399, and equation 2 can be applied across multiple devices using their respective neural network models described in abstract in page 1396. See algorithm 2 in page 1399 for more details.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of KIM J. and JAISWAL along with the teachings of MAO by using the teachings of KIM J. and JAISWAL of devices that obtain state information at time points, with MAO’s teaching of using an inference distribution ratio with an inference start point on electronic devices.
One of ordinary skill in the art would be motivated to do so because by integrating MAO’s framework into the methods of KIM J. and JAISWAL, one with ordinary skill in the art would achieve the goal of having a model that “can accelerate the DNN computation by 2.17-4.28×. Besides the parallel execution, the performance speedup also partially comes from the reduction of the data delivery time,” (MAO, page 1396, abstract), and “MoDNN also outperforms the conventional 2D-grids partition scheme by substantially reducing the data delivery time though the data transmission
size is slightly increased”, (MAO, page 1401, section D. Overall Evaluation of MoDNN).
Claim 6:
Regarding claim 6, KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 1.
Further, referring to claim 6, KIM J. teaches “The electronic device of claim 1, wherein the at least one processor is further configured to execute the at least one instruction to: … and transmit, via the transceiver,”
See KIM J. in paragraph [0140] describe “on the other hand, the server 100 can transmit not only the information about the error device but also the information about the determined operation status to the user terminal device. 4, the server 100 receives information indicating that an operation state in which cooling of the air conditioner 41 has been activated has occurred, and transmits information indicating that an error has occurred to the first temperature sensor 42, To the terminal device.” Here, the examiner construes transceiver to have functions both as a receiver and transmitter to carry and deliver information. Here, KIM J. mentions the server 100 has both receiver and transmitter roles which are included in the functions of a transceiver. See KIM J. in paragraph [0150] for more information.
Further, see KIM J. in paragraph [0165] discuss “meanwhile, the processor 130 of the server according to the embodiment of the present disclosure can determine whether there is an error in a plurality of sensors and electronic devices by using a learning algorithm, and provide information on a determination result to a user.” See KIM J. in paragraph [0058] mention “referring to FIG. 2, the server 100 includes a communication unit 110, a storage unit 120, and a processor 130.” KIM J. connects a processor with a transceiver (server 100). Here, KIM J. teaches at least one processor configured to execute at least one instruction. See KIM J. in paragraphs [0021-0024] for more information.
Further, MAO teaches “partition the artificial neural network according to the determined inference distribution ratio, and …, the partitioned artificial neural network to each of the plurality of devices corresponding to the determined inference distribution ratio,”
See MAO in abstract, page 1396 describes the model "MoDNN can partition already trained DNN models onto several mobile devices to accelerate DNN computations by alleviating device-level computing cost and memory usage." Here, MAO describes a model using deep neural networks or DNN (i.e. an artificial neural network) that partitions those models across multiple mobile devices (i.e. wherein the electronic device is determined from among the plurality of devices). In addition, MAO describes in page 1397 in section III. System Framework of MoDNN, first paragraph, and page 1400, section IV. Experiments, first paragraph for more information.
Later, see MAO in page 1397 section III, part A. Definition of Terminologies and Variables, MAO describes “Data Delivery Time: Data delivery time denotes the total time consumption for the data being transmitted between nodes. Data delivery time includes two parts: wakeup time and transmission time. Wakeup time represents the amount of time for the head of the data traveling from the sender to the receiver and transmission time denotes the amount of time for the receiver receiving from the first bit to the last bit of the data.” MAO in page 1396 describes a model using neural networks (same as an artificial neural network) that partitions those models across multiple mobile devices (i.e. wherein the electronic device is determined from among the plurality of devices). In page 1397, MAO is measuring two types of information for a first state, the first information is the wakeup time which is the elapsed time for data to travel from sender to the receiver, and a second information is the transmission time which is an elapsed time for receiver receiving data for each device. Both of these are considered to be a first state of a device. From the application’s specification in paragraph [0007], which states “each of the first state information and the second state information may include at least one of a usage rate of a central processing unit (CPU), a usage rate of a graphics processing unit (GPU), a temperature of the CPU, a temperature of the GPU, the number of executed applications, or an elapsed time of each of the plurality of devices”. Elapsed time is construed to mean any time that occurs regarding an operation of a device.
Further, see MAO in page 1399, section 3) Fine Grain Cross Partition (FGCP), paragraph 2, describe “In addition, FGCP needs to consider the discrepancy of execution time between the GO and the worker nodes during the offloading process, especially the data delivery time on the network between the worker node x and the GO, which can be conceptually formulated by:
PNG
media_image4.png
72
576
media_image4.png
Greyscale
where TPT is the mobile network throughput;” MAO describes the worder nodes to mean electronic devices, and each device gets an estimated time or variable Initial_ET for that device to implement a workload. The examiner construes inference distribution ratio as any mathematical operation that distributes computation or resource usage among various devices. Here, MAO describes that the first state is the delivery time incorporated into equation 2 (i.e. inference distribution ratio of the artificial neural network) with the numerator, worker node refers to a device, and TPT measures a throughput of mobile networks in page 1399, and equation 2 can be applied across multiple devices using their respective neural network models described in abstract in page 1396. Later, MAO describes in page 1400, section IV. Experiments, part A. Environment Setup and Testbench Selection that “The measured average WLAN wakeup time and transmission throughput are 54.7ms and 43.8Mbps, respectively.” TPT, which is a throughput measurement of mobile network, is interpreted as a measure of a device’s computation usage. Since each device is connected to a network, each device has their own throughput and computation usage (measured in page 1400 section IV in units Mbps), which is based on a second state. See MAO for more details in page 1398, section D. Weight Partition for Sparse FLs, first paragraph.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of KIM J. and JAISWAL along with the teachings of MAO by using the teachings of KIM J. and JAISWAL of devices that obtain state information at time points, with MAO’s teaching of using an inference distribution ratio on those devices with partitioned artificial neural network models.
One of ordinary skill in the art would be motivated to do so because by integrating MAO’s framework into the methods of KIM J. and JAISWAL, one with ordinary skill in the art would achieve the goal of having a model that “can accelerate the DNN computation by 2.17-4.28×. Besides the parallel execution, the performance speedup also partially comes from the reduction of the data delivery time,” (MAO, page 1396, abstract), and “MoDNN also outperforms the conventional 2D-grids partition scheme by substantially reducing the data delivery time though the data transmission
size is slightly increased”, (MAO, page 1401, section D. Overall Evaluation of MoDNN).
Claim 11:
Referring to claim 11, the claim recites similar limitations as corresponding claim 1 and is rejected for similar reasons as claim 1 using similar teachings and rationale.
Claim 15:
Regarding claim 15, the claim KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 11. Referring to claim 15, the claim recites similar limitations as corresponding claim 5 and is rejected for similar reasons as claim 5 using similar teachings and rationale.
Claim 16:
Regarding claim 16, the claim KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 11. Referring to claim 16, the claim recites similar limitations as corresponding claim 6 and is rejected for similar reasons as claim 6 using similar teachings and rationale.
Claim 20:
Regarding claim 20, KIM J. teaches “a non-transitory computer readable medium for storing computer readable program code or instructions which are executable by a processor to perform a method,”
See KIM J. in paragraph [0173] mention “when the data learning unit 1010 and the data recognition unit 1020 are implemented by a software module (or a program module including instructions), the software module may be a computer-readable and non-volatile readable recording medium non-transitory computer readable media. In this case, the software module may be provided by an operating system (OS) or by a predetermined application.” Later, see KIM J. describe in paragraph [0210] “a method of controlling a server according to various embodiments of the present disclosure described above may be stored in a non-transitory readable medium. Such non-transiently readable media can be used in various devices.” Here, KIM J. teaches a non-transitory computer readable medium used in various devices.
Further, KIM J. teaches “the method comprising: obtaining first state information at a first time point from each of devices comprising the electronic device;”
See KIM J. mention in paragraph [0051] “when the power of the TV 13 is turned on, the display of the TV is turned on, so that the first temperature / ambient light sensor 10 around the TV can detect a change in illuminance, Power consumption can be measured according to the supply... When an error occurs in the first temperature / ambient light sensor 10, when the TV 13 is turned on, the power consumption of the power measurement sensor 15 due to power supply to the TV is measured, and the first temperature / 10), … In other words, it can be seen that there is a problem in the first temperature / ambient light sensor 10 as compared with the case of the normal state.” Later, KIM J. notes that in paragraphs [0149-0150] “when the operation status of the at least one electronic device is determined after the new sensor is registered in the home network, the server 100 determines the sensing data received from the new sensor at the time point corresponding to the determined operation state, As the reference sensing data of the new sensor with respect to the operating conditions. The server 100 receives sensing data from a plurality of sensors in the home network (S920). The server 100 can continuously receive and store sensing data from a plurality of sensors. Alternatively, the server 100 may receive and store sensing data from a plurality of sensors at predetermined intervals.” Here, KIM J. describes measuring temperature from sensors at various time intervals, and detect measurements at various time points. Note, the examiner construes first to mean initial time selected. KIM J. also teaches first state information to be a temperature measurement of the device from first temperature in paragraph [0051], and measured at the time point corresponding to the determined operation state from [0149-0150] (i.e. any first time point). See KIM J. in paragraphs [0020] and [0028-0029] for more information.
However, KIM J. did not teach:
obtaining second state information of each of the devices at a second time point that is a preset time interval after the first time point, by inputting the first state information to a state inference model;
and determining an inference distribution ratio of an artificial neural network of each of the plurality of devices, based on the second information of each of the plurality of devices, wherein the electronic device is determined from among the plurality of devices, based on network states of the plurality of devices.
In an analogous system, JAISWAL teaches “obtaining second state information of each of the devices at a second time point that is a preset time interval after the first time point, by inputting the first state information to a state inference model;”
See JAISWAL in page 5, section V. Proposed Method, paragraph 3, describe "The model needs to know the input shape. For this reason, the first layer in a Sequential model needs to receive information about it. Input needs to be in the form [samples, time steps, features], where 'samples' is the total number of data points, 'time steps' is the number of time-dependent steps, 'features' refers to the number of variables." Here, JAISWAL shows the inputting of a time step (i.e. time point). See JAISWAL in page 6, V. Proposed Method, column 1, paragraph 3, describe "Rectified Linear Activation Function (ReLU) is used to build the TFE model as it allows it to learn nonlinear dependencies and complex mapping functions ... Where, Xt = input at current timestamp t
Yt = output at current timestamp t
Ht = output of LSTM block
Steps = 40 ", and refer to figures 9 and 10, and table IX for more details.
PNG
media_image2.png
240
704
media_image2.png
Greyscale
PNG
media_image3.png
892
1298
media_image3.png
Greyscale
Further, see JAISWAL describe in page 7, section VI. On-Device Results, first paragraph of “Apart from the flagship Samsung Galaxy S20, we have validated the TFE model across different Samsung smartphone devices having different form factor and internal designs, Samsung A50, A70 and Samsung Tablet Active 2.” Here, the time step mentioned by JAISWAL is interpreted by examiner to mean the time point. Here, JAISWAL shows that by inputting the first or initial elapsed time or first state information of Xt at current timestamp t, the temperature measurement (i.e. second state information) is obtained for each device (see table IX, where scenarios represent to each type of operation for various Samsung smartphone devices described in page 7, section VI ) at a second time point that is a preset time interval (from figure 9 shows set intervals of 30 seconds, 1 minute or 3 minutes) after the first time point.
Further, see JAISWAL in page 2, section II. Related Work, in second column, paragraphs 5-6, describe "certain limited exothermic components in the device have been utilized [10] to define the complex relationship between the various components inside of a mobile device and its surface temperature. Although comprehensive measurement results for a specific time window for predictions or for high load use cases like FastDischarge [15] and heavy high-end multiplayer games - PlayerUnknown's Battlegrounds (PUBG) [16] Counter - Strike: Global Offensive (CS GO) [17] may not have been calculated. Our prediction model is different from the above mentioned studies in that we use a complex Recurrent Neural Network - Long Short Term Memory, .... Also, ours is a multi-step model, which can be modified to accurately predict temperatures at multiple intervals at a time. While accurate predictions 5 seconds into the future have been made [13], the range of our prediction can be as long as 3 minutes, giving significant leverage to activate mitigation mechanisms well in advance." Here, JAISWAL describes predicting temperature, which is a state information, at multiple intervals at a time such as 5 seconds or 3 minutes into the future, and shows a specific time window or preset time interval after an initial or first time point.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the reference of KIM J. along with the teachings of JAISWAL by using the teachings of KIM J. of a method that uses a model for state inference, with JAISWAL’s teaching of using a device that obtains state information at time points.
One of ordinary skill in the art would be motivated to do so because by integrating JAISWAL’s framework into the system of KIM J., one with ordinary skill in the art would achieve the goal of providing a model that is “capable of providing accurate advance prediction of device surface temperature with average error rate of 0.64°C, which can efficiently handle temperature implications thereby enabling longer sustained device performance under 5G, gaming, multimedia and other heavy use cases. Clients can effectively use this thermal prediction to select the best available mitigation strategy,” (JAISWAL, page 7, section VII. Conclusion And Future Work).
However, KIM J. in view of JAISWAL did not teach “and determining an inference distribution ratio of an artificial neural network of each of the plurality of devices, based on the second information of each of the plurality of devices, wherein the electronic device is determined from among the plurality of devices, based on network states of the plurality of devices,”
In an analogous art, MAO teaches “and determining an inference distribution ratio of an artificial neural network of each of the plurality of devices, based on the second information of each of the plurality of devices, wherein the electronic device is determined from among the plurality of devices, based on network states of the plurality of devices,”
See MAO in abstract, page 1396 describes the model "MoDNN can partition already trained DNN models onto several mobile devices to accelerate DNN computations by alleviating device-level computing cost and memory usage." Here, MAO describes a model using deep neural networks or DNN (i.e. an artificial neural network) that partitions those models across multiple mobile devices (i.e. wherein the electronic device is determined from among the plurality of devices). In addition, MAO describes in page 1397 in section III. System Framework of MoDNN, first paragraph, and page 1400, section IV. Experiments, first paragraph for more information.
Later, see MAO in page 1397 section III, part A. Definition of Terminologies and Variables, MAO describes “Data Delivery Time: Data delivery time denotes the total time consumption for the data being transmitted between nodes. Data delivery time includes two parts: wakeup time and transmission time. Wakeup time represents the amount of time for the head of the data traveling from the sender to the receiver and transmission time denotes the amount of time for the receiver receiving from the first bit to the last bit of the data.” Here in page 1397, MAO is measuring two types of information for a first state, the first information is the wakeup time which is the elapsed time for data to travel from sender to the receiver, and a second information is the transmission time which is an elapsed time for receiver receiving data for each device. Both of these are considered to be a first state of a device. From the application’s specification in paragraph [0007], which states “each of the first state information and the second state information may include at least one of a usage rate of a central processing unit (CPU), a usage rate of a graphics processing unit (GPU), a temperature of the CPU, a temperature of the GPU, the number of executed applications, or an elapsed time of each of the plurality of devices”. Elapsed time is construed to mean any time that occurs regarding an operation of a device.
Further, see MAO in page 1399, section 3) Fine Grain Cross Partition (FGCP), paragraph 2, describe “In addition, FGCP needs to consider the discrepancy of execution time between the GO and the worker nodes during the offloading process, especially the data delivery time on the network between the worker node x and the GO, which can be conceptually formulated by:
PNG
media_image4.png
72
576
media_image4.png
Greyscale
where TPT is the mobile network throughput;” The examiner construes inference distribution ratio as any mathematical operation that distributes computation or resource usage among various devices. Here, MAO describes that the first state is the delivery time incorporated into equation 2 (i.e. inference distribution ratio of the artificial neural network) with the numerator, worker node refers to a device, and TPT measures a throughput of mobile networks in page 1399, and equation 2 can be applied across multiple devices using their respective neural network models described in abstract in page 1396. See algorithm 2 in page 1399 for more details.
Later, MAO describes in page 1400, section IV. Experiments, part A. Environment Setup and Testbench Selection that “The measured average WLAN wakeup time and transmission throughput are 54.7ms and 43.8Mbps, respectively.” TPT, which is a throughput measurement of mobile network, is interpreted as a measure of a device’s computation usage. Since each device is connected to a network, each device has their own throughput and computation usage (measured in page 1400 section IV in units Mbps), which is based on a second state.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of KIM J. and JAISWAL along with the teachings of MAO by using the teachings of KIM J. and JAISWAL of devices that obtain state information at time points, with MAO’s teaching of using an inference distribution ratio with state information on electronic devices.
One of ordinary skill in the art would be motivated to do so because by integrating MAO’s framework into the methods of KIM J. and JAISWAL, one with ordinary skill in the art would achieve the goal of having a model that “can accelerate the DNN computation by 2.17-4.28×. Besides the parallel execution, the performance speedup also partially comes from the reduction of the data delivery time,” (MAO, page 1396, abstract).
Claims 2 and 12 are rejected under 35 U.S.C. 103 in view of KIM J., further in view of JAISWAL, further in view of MAO, and further in view of Shindo T. et al. (Pub. No. JP 5443686-B), published on March 19, 2014, (hereafter, SHINDO).
Claim 2:
Regarding claim 2, KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 1.
However, KIM J. in view of JAISWAL, further in view of MAO, did not explicitly teach “the electronic device of claim 1, wherein “each of the first state information and the second state information comprises at least one of a usage rate of a central processing unit (CPU), a usage rate of a graphics processing unit (GPU), a temperature of the CPU, a temperature of the GPU, the number of executed applications, or an elapsed time of each of the plurality of devices,”
In an analogous system, SHINDO teaches “the electronic device of claim 1, wherein each of the first state information and the second state information comprises at least one of a usage rate of a central processing unit (CPU), a usage rate of a graphics processing unit (GPU), a temperature of the CPU, a temperature of the GPU, the number of executed applications, or an elapsed time of each of the plurality of devices,”
See SHINDO in paragraph [0043], where SHINDO describes "load amount calculation process required for the entire system to operate the service A will be specifically described with reference to FIG. For example, when the determination unit 104 recognizes that there are two servers that provide the service A from the acquired current configuration information and acquires the event information 1 to 3 shown in FIG. CPU utilization before 30 minutes, current CPU utilization is 50%, CPU utilization after 30 minutes is 70%, CPU utilization after 60 minutes is 90%, CPU utilization after 90 minutes is 110% Therefore, as shown in FIG. 9, the total of the CPU performance ratio × CPU usage rate for two units is calculated. Here, the case where both the AP servers 1 and 2 have a CPU performance ratio of 1 is exemplified, " Here, SHINDO teaches using a ratio equation to calculate CPU performance ratio (i.e. inference distribution ratio) × CPU usage rate among two servers, which can be from two devices. The second state SHINDO shows can be from either CPU or elapsed time, which is seen as number of minutes required to operate a service. SHINDO teaches a CPU usage rate as a first state and an elapsed time in minutes as a second state information, which overall SHINDO teaches the first state information and the second state information. From the specification paragraph [0045], the language A or B can be interpreted as both or either or. See SHINDO in paragraphs 0025, 0032, 0035-37, or 0044 for more information.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of KIM J., JAISWAL, and MAO, along with the teachings of SHINDO by using the teachings of KIM J., JAISWAL, MAO, of a device that obtains state information at time points, with SHINDO’s teaching of a first state information and a second state information.
One of ordinary skill in the art would be motivated to do so because by integrating SHINDO’s framework into the methods of KIM J., JAISWAL, MAO, , one with ordinary skill in the art would achieve the goal of providing an “autonomous computing system is known in which a computer manages itself in order to reduce the burden of managing a computer system by humans” (SHINDO, paragraph [0002]), and “is actually expressed in a programming language so that the computer can understand it,” (SHINDO, paragraph [0041]).
Claim 12:
Regarding claim 12, the claim KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 11. Referring to claim 12, the claim recites similar limitations as corresponding claim 2 and is rejected for similar reasons as claim 2 using similar teachings and rationale.
Claims 3 and 13 are rejected under 35 U.S.C. 103 over KIM J., further in view of JAISWAL, further in view of MAO, further in view of SHINDO, and further in view of Schmidt M. et al. (Pub. No. WO 2020251059-A1), published on December 17, 2020, (hereafter, SCHMIDT).
Claim 3:
Regarding claim 3, KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 1.
However, KIM J. in view of JAISWAL, further in view of MAO, did not teach
“normalize an inverse number of the elapsed time of each of the plurality of devices,”
"determine the normalized inverse number of the elapsed time, as the inference distribution ratio of the artificial neural network of each of the plurality of devices,"
In an analogous system, SHINDO teaches “normalize an inverse number of the elapsed time of each of the plurality of devices,”
See SHINDO in paragraphs [0045] mentions “FIG. 10 is a diagram illustrating a method for calculating the necessary number of servers by the determination unit 104. As shown in FIG. 10, the required number of servers is obtained by dividing the load amount of the entire computer system required for providing the service A by the CPU utilization rate per one 60% indicated in the policy. Is calculated. The number of servers required at present is calculated as 2, the number of servers required after 30 minutes is 3, the number of servers required after 60 minutes is calculated, and the number of servers required after 90 minutes is calculated as 4.”
Further, SHINDO explains in paragraph [0050] that “ the event information may be converted into CBE (Common Base Event) data, which is one of the standard formats, in implementing the present invention.” Note, normalize is construed to mean to process data so data will be bringing data, processes, or behaviors into conformity with an established standard, norm, or scale, and can also mean standardize or follow a standard. Here, SHINDO describes in paragraphs [0045 and 0050] and in figure 10 that the whole data is converted to the standard format of common base event data, and SHINDO shows an inverse of the elapsed time of each of the devices in the calculation from a load of one service on a device in figures 10 and 11. See SHINDO describe in paragraphs [0037, 0043-44] for more information.
PNG
media_image6.png
766
510
media_image6.png
Greyscale
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of KIM J., JAISWAL, and MAO, along with the teachings of SHINDO by using the teachings of KIM J., JAISWAL, MAO, of a device that obtains state information at time points, with SHINDO’s teaching of normalize an inverse number of the elapsed time of each of the devices from the state information.
One of ordinary skill in the art would be motivated to do so because by integrating SHINDO’s framework into the methods of KIM J., JAISWAL, MAO, one with ordinary skill in the art would achieve the goal of providing an “autonomous computing system is known in which a computer manages itself in order to reduce the burden of managing a computer system by humans” (SHINDO, paragraph [0002]), and “is actually expressed in a programming language so that the computer can understand it,” (SHINDO, paragraph [0041]).
However, KIM J. in view of JAISWAL, further in view of MAO, and further in view of SHINDO did not teach "determine the normalized inverse number of the elapsed time, as the inference distribution ratio of the artificial neural network of each of the plurality of devices,"
In an analogous system, SCHMIDT teaches "determine the normalized inverse number of the elapsed time, as the inference distribution ratio of the artificial neural network of each of the plurality of devices,"
See SCHMIDT in paragraph [0045] describe, "At S1054, an applying section applies the normalizing flow model to the random sample of time series data to generate an artificial normal sample of time series data. More specifically, the applying section feeds the random sample of time series data inversely through the normalizing flow model to output an artificial normal sample of time series data that has substantially the same probability distribution as the real normal sample of time series data." Note the examiner has construed normalized to mean processing data to be standardized or normalized similar to a statistics probability distribution. Here, SCHMIDT shows that inputting the time series data inversely through the normalizing flow model to output a normal sample of time series data (i.e. determine the normalized inverse number of the elapsed time ) is seen as part of the inference to the model and part of the inference distribution ratio.
Further, see SCHMIDT in paragraph [0011] describe “normalizing flows are a class of deep generative models that may leverage invertible neural networks to learn a mapping between a simple base distribution and a given data distribution,” and see SCHMIDT in paragraph [0065] mention “many of the embodiments of the present invention include artificial intelligence, learning processes, and neural networks in particular. Some of the foregoing embodiments describe specific types of neural networks.” Note the examiner construes artificial neural network to have similar meaning as a neural network. Here, SCHMIDT shows that this process is applied in neural network models.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of KIM J., JAISWAL, MAO, and SHINDO along with the teachings of SCHMIDT by using the teachings of KIM J., JAISWAL, MAO, and SHINDO of a device that obtains state information at time points of a normalized inverse of the elapsed time, with SCHMIDT’s teaching of the process of determining the normalized inverse number of the elapsed time, as the inference distribution ratio for each of the devices in a neural network model.
One of ordinary skill in the art would be motivated to do so because by integrating SCHMIDT’s framework into the methods of KIM J., JAISWAL, MAO, and SHINDO, one with ordinary skill in the art would achieve the goal of providing “normalizing flow models are generative models which may produce tractable distributions where both sampling and density evaluation can be efficient and exact, whereas many other density estimation models can do only one or another, and which may learn very complex probability distributions. Normalizing flow models with an autoregressive property may be beneficial for any sequential data where there exists some relationship between two adjacent data points, includes time series data.” (SCHMIDT, paragraph [0060]).
Claim 13:
Regarding claim 13, the claim KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 11. Referring to claim 13, the claim recites similar limitations as corresponding claim 3 and is rejected for similar reasons as claim 3 using similar teachings and rationale.
Claims 4 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over KIM J. in view of JAISWAL, further in view of MAO, further in view of Sikder A. et al., in “A Context-Aware Framework for Detecting Sensor-Based Threats on Smart Devices,” available at https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8613866, published on January 16, 2019, (hereafter, SIKDER).
Claim 4:
Regarding claim 4, KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 1.
Further, KIM J. teaches “the electronic device of claim 1, wherein the at least one processor is further configured to execute the at least one instruction,”
See KIM J. in paragraph [0071] mention "the server according to any of (1)-(3), wherein the processing circuitry is further configured to compare the received performance data of the device to the stored other performance data, wherein the device is one of a mobile communication device, a wireless user equipment, a smart phone, wearable technology, and a tablet computer, an embedded processor." Here, KIM J. teaches at least one processor configured to execute the at least one instruction.
However, referring to claim 4, KIM J. in view of JAISWAL did not teach
“to: obtain third state information comprising at least one of whether a preset application is executed, whether a screen is turned on, or whether a camera is executed, at the first time point,
“and obtain the second state information based on additionally inputting the third state information to the state inference model.”
In an analogous system, SIKDER teaches “to: obtain third state information comprising at least one of whether a preset application is executed, whether a screen is turned on, or whether a camera is executed, at the first time point,”
See page 251, section 5.3: Other ML-based techniques where SIKDER describes that a "neural network is another common technique that is utilized by researchers for malware detection. In neural network techniques, the relation between attributes of dataset is compared with the biological neurons and a relation map is created to observe the changes for each attribute [49]. We chose Multilayer Perceptron algorithm for training the 6thSense framework as it can distinguish relationships among non-linear dataset." Here, this shows that multilayer perceptron is a type of neural network algorithm used for the model.
Further, see pages 251-252, section 6.1, Data Collection Phase, SIKDER describes "Permission-Imposed Sensors in 6thSense. We chose five permission-imposed sensors to build the context-aware model (microphone, GPS, speaker, camera, and headset) of 6thSense. The conditions of these sensors can be represented by their logical states (on/off status) for different user activities." Further, see SIKDER mention in section 6.1 Data Collection Phase, page 252 that "in summary, the aforementioned App collects data from eight different sensors for different typical user activities. 6thSense observes sensor state (combination of working conditions (i.e., values, on/off status) of different sensors) in a per second manner for each user activity. " Here, SIKDER shows to obtain third state information involves at least one of the types of information of whether a screen or camera is turned on, per second (i.e. at the first time point). Note the examiner construes third state information with any of the following types of information from the specification in paragraph [0019], note “the obtaining of the second state information comprises: obtaining third state information comprising at least one of whether a preset application is executed, whether a screen is turned on, or whether a camera is executed,” and in specification paragraph [0017] notes the definition of a second state information in “each of the first state information and the second state information comprises at least one of a usage rate of a central processing unit (CPU), a usage rate of a graphics processing unit (GPU), a temperature of the CPU, a temperature of the GPU, the number of executed applications, or an elapsed time of each of the plurality of devices.”
Further, SIKDER teaches “and obtain the second state information based on additionally inputting the third state information to the state inference model,”
See SIKDER in page 248, section 4, Adversary Model and Assumptions, where SIKDER talks about "6thSense divides the total execution time of an activity into smaller times and observes the sensors’ states (on/off) over a short time span. Thus, whenever a sensor state is changed, 6thSense can understand the context and take a decision according to the context. For example, while a user is walking with a smartphone on his hand, several sensors (i.e., accelerometer, gyroscope, light sensor, etc.) remain active. If we divide the time of the activity in smaller times, we can see different sets of sensors active for different sensor states (Fig. 2). In this way, 6thSense considers all device states to understand the context of the activity and differentiate between benign and malicious activities." Here, SIKDER mentions the model considers all device states and incorporates this information into the model. Specifically, SIKDER explicitly teaches a state being a sensor or camera that can be turned on or off over a time span, which shows trying to observe an elapsed time (i.e. second state information) based on inputting sensors’ states (on/off) (i.e. third state information) to the model.
Further, see page 257, section 7.7, Performance Overhead, where SIKDER describes "6thSense collects data in an Android device from different sensors (permission and no-permission imposed sensors). In this sub-section, we measure the performance overhead introduced by 6thSense on the tested Android devices (smart watch and smartphone) in terms of CPU usage, RAM usage, file size, and power consumption. Tables 7, 8, and 9 give the details of the performance overhead of 6thSense on the smart watch and the smartphone." Here, SIKDER shows that the CPU usage information (i.e. second state information) is also obtained while inputting the sensors’ on/off state (i.e. third state information) for the model. Overall, SIKDER teaches obtain the second state information based on additionally inputting the third state information to the state inference model.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of KIM J., JAISWAL, and MAO, along with the teachings of SIKDER by using the teachings of KIM J., JAISWAL, and MAO, of a device that obtains state information at time points for a model, with SIKDER’s teaching of that state inference model has a third state information includes a screen or camera turned on or off.
One of ordinary skill in the art would be motivated to do so because by integrating SIKDER’s framework into the systems of KIM J., JAISWAL, and MAO, one with ordinary skill in the art would achieve the goal of providing a model that “is highly effective and efficient at detecting sensor-based attacks while yielding minimal overhead”, (SIKDER, page 259, section 8. Conclusion).
Claim 14:
Regarding claim 14, the claim KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 11. Referring to claim 14, the claim recites similar limitations as corresponding claim 4 and is rejected for similar reasons as claim 4 using similar teachings and rationale.
Claims 7 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over KIM J. in view of JAISWAL, further in view of MAO, further in view of Jung N. et al., (Pub. No. DE 112020003498 T5), published on April 21, 2022, (hereafter, JUNG).
Claim 7:
Regarding claim 7, KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 1.
Further, KIM J. in view of JAISWAL, further in view of MAO, teaches “and target state information at a fourth time point after a preset time interval from the third time point,”
See JAISWAL in page 2, section II. Related Work, in second column, paragraphs 5-6, describe “also, ours is a multi-step model, which can be modified to accurately predict temperatures at multiple intervals at a time. While accurate predictions 5 seconds into the future have been made [13], the range of our prediction can be as long as 3 minutes,” and in abstract mention in page 5, section V. Proposed Method “as above section showcase that LSTM model giving high accuracy on the time series data we are using to predict the multiple instances of time at once.” Note the examiner construes third or fourth time points to be subsequent time points from an initial timepoint that is determined by the person running the model experiments. Here, JAISWAL shows that this model they used in their study is able to predict temperatures (which is a state information) at multiple intervals at a time, with a range of predictions from an interval of 5 seconds to 3 minutes into the future, (which relates to a preset time interval). Here, JAISWAL is teaching the ability to predict an input temperature (state information) at a future timepoint (relates to fourth timepoint) after a preset time interval from the previous time point (i.e. third time point).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the reference of KIM J. along with the teachings of JAISWAL by using the teachings of KIM J. of a device with a processor and memory that uses artificial neural network models, with JAISWAL’s teaching of using a device that obtains state information at time points.
One of ordinary skill in the art would be motivated to do so because by integrating JAISWAL’s framework into the system of KIM J., one with ordinary skill in the art would achieve the goal of providing a model that is “capable of providing accurate advance prediction of device surface temperature with average error rate of 0.64°C, which can efficiently handle temperature implications thereby enabling longer sustained device performance under 5G, gaming, multimedia and other heavy use cases. Clients can effectively use this thermal prediction to select the best available mitigation strategy,” (JAISWAL, page 7, section VII. Conclusion And Future Work).
However, KIM J. in view of JAISWAL, further in view of MAO, did not explicitly teach “the electronic device of claim 1, wherein the state inference model is regression-trained based on an input of state information for training at a third time point and target state information at a fourth time point after a preset time interval from the third time point.”
In an analogous system, JUNG teaches “the electronic device of claim 1, wherein the state inference model is regression-trained based on an input of state information for training at a third time point,”
See paragraph [0023] where JUNG talks about "In some configurations, the processor 170 includes at least one inference module 172 and one training data control module 174. Generally, inference module 172 uses a neural network to perform classification or regression results, and training data control module 174 performs functions that learn from the training data and train the neural network to perform specific tasks". Here, JUNG in [0023] teaches that the inference model is regression-trained or can train to perform tasks such as regression or classification.
Further, see JUNG in paragraph [0049], describe “supervised learning is a method of machine learning in which a single function is derived from the training data. Among the functions that can be derived in this way, a function that outputs a continuous range of values can be called a regressor, and a function that predicts and outputs the class of an input vector can be called a classifier. In supervised learning, a CNN can be trained with training data that has been assigned a label. The label can refer to a target response (or result value) that can be guessed by the artificial neural network when the training data is entered into the artificial neural network. In this description, the target response (or a result value) to be guessed by CNN when entering the training data can be referred to as label or labeling data.” Here, JUNG notes that inference on state information can include using an artificial neural network model.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of KIM J., JAISWAL, and MAO, along with the teachings of JUNG by using the teachings of KIM J., JAISWAL, and MAO, of a device that obtains state information at time points for a model, with JUNG’s teaching of that state inference model is regression-trained.
One of ordinary skill in the art would be motivated to do so because by integrating JUNG’s framework into the systems of KIM J., JAISWAL, and MAO, one with ordinary skill in the art would achieve the goal of providing “executing the various processes at different stages is advantageous because it helps the Edge Device 150 to use its limited hardware resources more efficiently,” (JUNG, paragraph [0026]).
Claim 17:
Regarding claim 17, the claim KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 11. Referring to claim 17, the claim recites similar limitations as corresponding claim 7 and is rejected for similar reasons as claim 7 using similar teachings and rationale.
Claims 8 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over KIM J. in view of JAISWAL, further in view of MAO, further in view of Najari, N. et al. in "Network Traffic Modeling For IoT-device Re-identification," published for a conference August 31, 2020 - September 2, 2020, available on https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9191376, (hereafter, NAJARI).
Claim 8:
Regarding claim 8, KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 1.
However, KIM J. in view of JAISWAL, further in view of MAO, did not teach “the electronic device of claim 1, wherein the network states are network input/output (I/O) packet amounts of the plurality of devices based on test information received by a first device from among the plurality of devices excluding the first device, the first device being randomly selected from the plurality of devices.”
In an analogous system, NAJARI teaches “the electronic device of claim 1, wherein the network states are network input/output (I/O) packet amounts of the plurality of devices based on test information received by a first device from among the plurality of devices excluding the first device, the first device being randomly selected from the plurality of devices,”
See NAJARI in page 3, section A. Feature Extraction And Preprocessing, fourth paragraph, describe "we propose to pre-process not only the network packets sent by the device, but also the received packets. While sent packets reflect device-inherent characteristics (e.g. capacity, memory), received packets indicate the impact of an external configuration or interaction with other appliances on the device behavior. As we use a bidirectional flow of network traces..." Here, NAJARI describes input and output network packets with input correspond to receiving, and output correspond to sending network packets (along with characteristics such as capacity or amount contained in the network packet), illustrating the bidirectional flow of network packets.
Further, see NAJARI in page 4, section IV. Experimental Results, section B. Training and testing strategies, paragraph 2, describe "use Time-series Cross-Validation (TCV). Indeed, the use of traditional cross-validation causes data leakage because this approach assumes that the samples are independent and data are split randomly [22]. Since pre-processed packets are chronologically linked and dependent on each other, we need to split data temporally where the test subset is chronologically after the training one i.e. we train the model on past observations and we predict future ones. TCV involves an outer loop that splits the dataset into multiple different training and test sets". Here, NAJARI teaches using test data. See NAJARI in page 2, section III. Proposed Approach, second paragraph for more detail.
See NAJARI also in page 1, section I. Introduction, paragraph 3, mention "However, relying only on these parameters to identify IoT devices is not consistent for many reasons. First, declared MAC and IP addresses can be modified. This approach, a.k.a. address spoofing, is used not only for malicious purposes (usurpation) but also for privacy concerns (randomization) to prevent tracking and protect personal data." Here, the examiner notes that since MAC and IP addresses are part of network packet data, NAJARI teaches that the device associated with the network packet can also be randomized to protect personal data and not link certain information to a specific device.
Further, see NAJARI in page 2, section III. Proposed Approach, first paragraph, describe "Thus, the proposed approach relies only on network packet metadata and payload length for the identification. In addition, with the continuous growth of the IoT market, new devices need to be identified and integrated into the IoT platform. We need to dynamically add a new device or remove an existing one from the LAN." Note the examiner has construed first to mean an initial or one network packet selected initially, or to start with, and after selecting an initial device, other subsequent devices also can be selected later on. Here, NAJARI teaches dynamically adding a new device or removing an existing device from the network, which relates to test information received by a first device, then later excluding the first device. See NAJARI in page 3, section A. Feature Extraction And Preprocessing, first paragraph for more information.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of KIM J., JAISWAL, and MAO, by using the teachings of KIM J., JAISWAL, and MAO, of a device that obtains state information at time points for inferencing in a model, with NAJARI’s teaching of a selected candidate device connected to a wired network and has a network I/O packet amount equal to or smaller than a preset packet amount.
One of ordinary skill in the art would be motivated to do so because by integrating NAJARI’s framework into the methods of KIM J., JAISWAL, and MAO, one with ordinary skill in the art would achieve the goal of providing a method that can have a "main advantage of using generative models remains in their ability to learn temporal correlations in data and to forecast future events," (See NAJARI in page 2, paragraph 2, section III. Proposed Approach), with a method that "can detect repeatable patterns of network traffic traces and model device behavior, " (See NAJARI, page 3, section B. Machine Learning Models, part 1) Hidden Markov Models, last paragraph in section before part 2) RNN-LSTM), as well as "show that this approach improves the state-of-the-art of OS discovery and leads to a more accurate and less intrusive method,” (see NAJARI, page 2, paragraph 2, section II. Related Work, section C. Hybrid Device Fingerprinting).
Claim 18:
Regarding claim 18, the claim KIM J. in view of JAISWAL, further in view of MAO, teaches the limitations of claim 11. Referring to claim 18, the claim recites similar limitations as corresponding claim 8 and is rejected for similar reasons as claim 8 using similar teachings and rationale.
Claims 9 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over KIM J. in view of JAISWAL, further in view of MAO, further in view of NAJARI, further in view of Yang S. et al. (US PG PUB. No. US20150146523 A1), published on May 28, 2015, (hereafter, YANG).
Claim 9:
Regarding claim 9, KIM J. in view of JAISWAL, further in view of MAO, and further in view of NAJARI, teaches the limitations of claim 8.
However, KIM J. in view of JAISWAL, further in view of MAO, and further in view of NAJARI did not teach “the electronic device of claim 8, wherein the electronic device is a candidate device connected to a wired network from among at least one candidate device that is selected from among the plurality of devices and has a network I/O packet amount equal to or smaller than a preset packet amount.”
In an analogous art, YANG teaches “the electronic device of claim 8, wherein the electronic device is a candidate device connected to a wired network from among at least one candidate device that is selected from among the plurality of devices and has a network I/O packet amount equal to or smaller than a preset packet amount,”
See YANG in paragraph [0038] describe " The method of FIG. 5 is similar to that of FIG. 4, a difference between the two is that, a number of packets are processed each time in the method of FIG. 4, and only one packet is processed each time in the method of FIG. 5. The method of FIG. 5 is suitable for the multiple-interface network device having a smaller memory." See YANG in paragraph [0003] mention “the disclosure relates to a multiple-interface network device, and more particularly, to a selection method for transmitting network packets for a multiple-interface network device. Here, YANG shows that the method applies for selecting a device in the network.
Further, see YANG in paragraphs [0036-0037] describe "If the proportion of the packet amount of each of the traffic pattern types is less than the preset value, the traffic pattern type of the service flow may be set to a preset traffic pattern type among the traffic pattern types. For instance, it is assumed that the preset value is 90%, and the preset traffic pattern type is the bursting traffic pattern. If the proportion of the packet amount belonging to the periodical traffic pattern in the traffic log reaches 90%, the service flow belongs to the periodical traffic pattern. If the proportion of the packet amount of each traffic pattern in the traffic log is less than 90%, the service flow belongs to the bursting traffic pattern.
In step 450 of another embodiment, the preset value being less than or equal to 50% (e.g., 40%) may be set. Further, each of the traffic pattern types in the traffic pattern type lookup list may include different priorities." Note the examiner construes a preset packet amount to be the number of packets that gets transmitted or received by a device over time, and can represent any amount integer or percentage to quantify this amount. Here in [0036-0037], YANG describes a service flow rate of traffic pattern of a packet being processed, and listed some numbers as 40% or 50%. This rate is defined by YANG in [0036] to be “proportion of the packet amount of each of the traffic pattern types”, and includes a preset amount in packet transmission for devices. See YANG in paragraph [0035] for more information on the transmission amounts per packet for a device.
Further, see YANG in paragraph [0017] describe “Each of the network interfaces 140 is connected to one of different communication networks, such as Wi-Fi network, LTE network, Ethernet or other similar wired or wireless networks.” YANG here shows this applies to a wired network as well as to wireless networks.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of KIM J., JAISWAL, MAO, and NAJARI along with the teachings of YANG by using the teachings of KIM J., JAISWAL, MAO, and NAJARI of a device that obtains state information at time points for inferencing in a model, with YANG’s teaching of a selected candidate device connected to a wired network and has a network I/O packet amount equal to or smaller than a preset packet amount.
One of ordinary skill in the art would be motivated to do so because by integrating YANG’s framework into the methods of KIM J., JAISWAL, MAO, and NAJARI, one with ordinary skill in the art would achieve providing a method that can let “multiple-interface network device and the selection method for transmitting network packets may provide high efficiency in satisfying requirements of quality of service in guaranteed bit rate, thereby improving user experiences,” (YANG, paragraph [0040]).
Claim 19:
Regarding claim 19, the claim KIM J. in view of JAISWAL, further in view of MAO, and further in view of NAJARI, teaches the limitations of claim 18. Referring to claim 19, the claim recites similar limitations as corresponding claim 9 and is rejected for similar reasons as claim 9 using similar teachings and rationale.
Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over KIM J. in view of JAISWAL, further in view of MAO, further in view of NAJARI, further in view of YANG, and further in view of SURESH A. et al, (Pub. No. AU2018301289 B2), published on July 1, 2021, (hereafter, SURESH).
Claim 10:
Regarding claim 10, KIM J. in view of JAISWAL, further in view of MAO, further in view of NAJARI, and further in view of YANG, teaches the limitations of claim 9.
However, KIM J. in view of JAISWAL, further in view of MAO, further in view of NAJARI, and further in view of YANG, fail to teach “the electronic device of claim 9, wherein the electronic device is a candidate device having a highest GPU throughput from among the at least one candidate device,”
In an analogous art, SURESH teaches “the electronic device of claim 9, wherein the electronic device is a candidate device having a highest GPU throughput from among the at least one candidate device,”
See SURESH in paragraph [0003] describe "A computing device, such as a server or a desktop computing device, may rank any available GPUs based on their performance and computational capacity. Subsequently, the computing device may default to a highest ranked GPU for performing visual renderings..." Note the examiner construes GPU throughput to mean the number of operations a GPU can complete per unit of time. Here, SURESH describes ranking a device by the highest GPU, which shows the device choses a high performing GPU to process visual-based tasks in this case.
Later, SURESH notes in paragraph [0009] that "...the virtual GPU manager may enumerate each of the plurality of physical GPUs to identify a number of available physical GPUs. The virtual GPU manager may classify each of the available physical GPUs based on the processing capacity of each of the available physical GPUs. Responsive to classifying each of the available physical GPUs, the virtual GPU manager may rank each of the available physical GPUs based on the processing capacity." Here, SURESH mentions that the system can identify GPU by the processing capacity, similar to processing number of tasks per time.
Further, see SURESH in paragraph [0077] describe "virtual GPU manager 522 may distribute processing power equally across each of the super-GPU views or, alternatively, may allocate the available physical GPUs between the super-GPU views based the processing performance variables in a task specific manner." Later, see SURESH in paragraph [0071] illustrates "as such, the processing performance variables (e.g., power demand, processing bandwidth, processing capacity, floating point operations per second, render output units, texture units, texture fill-rate, pixel fill-rate, base frequency, boost frequency, memory clock rate, memory capacity, memory bandwidth, and the like) may vary across one or more of integrated CPU/GPU(s) 512 and/or one or more of discreet GPUs 514A-514N. In some instances, each of the one or more integrated CPU/GPU(s) 512 and/or one or more discreet GPUs 514A 514N may be used by computing device 501 for general purpose computational processing and/or graphical processing." Here, SURESH describes that processing performance is evaluated by a task specific manner, and in [0071] elaborates the types of performance variables that can be used. For more details, see SURESH in paragraphs [0075, 0076, 0115]
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the references of KIM J., JAISWAL, and MAO, NAJARI, YANG, along with the teachings of SURESH by using the teachings of KIM J., JAISWAL, MAO, NAJARI, and YANG, of a device that obtains state information at time points for inferencing in a model, with SURESH’s teaching of a candidate device having a highest GPU.
One of ordinary skill in the art would be motivated to do so because by integrating SURESH’s framework into the methods of KIM J., JAISWAL, MAO, NAJARI, and YANG, one with ordinary skill in the art would achieve the goal of providing " graphical rendering requests may be distributed across each of the available physical GPUs in the logical linkage, thereby aggregating the processing power of the summation of available physical GPUs into a single logical object. In this way, during performance of rendering requests, each of the available physical GPUs are actively leveraged to perform the computations corresponding to the rendering request, as opposed to conventional systems in which only the processing capacity of a most powerful available physical GPU is harnessed," (SURESH, paragraph [0076]).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to WENWEI ZENG whose telephone number is (571)272-7111. The examiner can normally be reached Monday-Friday, 8am-5pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Usmaan Saeed can be reached at (571) 272-4046. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/WenWei Zeng/Examiner, Art Unit 2146
/USMAAN SAEED/Supervisory Patent Examiner, Art Unit 2146