DETAILED ACTION
Claims 1-20 are pending in the Instant Application.
Claims 1-20 are rejected (Final Rejection).
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 .
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-20 are rejected under 35 U.S.C. 102(a)(2) as being unpatentable by UCHIYAMA et al.(“Uchiyama”), United States Patent Application Publication No. 2024/0259464.
As per claim 1, Uchiyama discloses a method comprising:
generating an intermediate representation of a neural network, wherein the neural network is comprised of a plurality of layers ([0079] and [0080] wherein the neural network is split according to layers, each layer having different output data sizes and computation delay, the output of the layer being the intermediate representation);
extracting, via a profiler of a splitter device, a plurality of neural network features based on the intermediate representation ([0080] and [0125] wherein for each layer, a computational delay is estimated and output size is determined);
selecting, via a classifier of the splitter device, one or more heuristics based on the neural network features ([0141]-[0143] wherein the splitter can choose between one or more heuristics by selecting one or more splitting modes);
determining one or more device characteristics of the one or more devices, wherein the one or more devices are connected to the splitter device ([0101] wherein the device capability (computation handler acting as the devices in the prior art) is determined);
determining a plurality of slices ([0141] wherein the slices (recognized as the range of layers) is determined) based on the neural network features, the one or more heuristics and the device characteristics ([0141] wherein the neural network features (execution delay of the ML in the prior art), the heuristics (a particular switching mode), and the device characteristics (load of the devices in the prior art) are used determine what layers are in each slice as shown in [Fig. 11])wherein each slice of the plurality of slices is associated with at least one of the devices, and
wherein each of the plurality of slices is associated with one or more of the plurality of layers ([0162] wherein slices (sets of layers in the prior art) are assigned to devices (nodes in the prior art)); and
generating the plurality of slices ([0079] and [0187] wherein the ranges of layers is distributed and generated on the different devices).
As per claim 2, Uchiyama discloses the method of claim 1 further comprising:
transmitting each of the plurality of slices to the associated device of the one or more devices ([0187] wherein distribution of different ranges for the slices is described).
As per claim 3, Uchiyama discloses the method of claim 1, wherein the classifier is comprised of a classifier neural network ([Fig. 2] wherein an image of a cat is classified as a cat).
As per claim 4, Uchiyama discloses the method of claim 1, wherein determining one or more device characteristics of the one or more devices comprising:
querying the one or more connected devices; and receiving, based on the query, the device characteristics ([0120] wherein the instructions to acquire the device characteristic (such as capability and load in the prior art).are the query, and the information is received in return).
As per claim 5, Uchiyama discloses the method of claim 1, wherein the one or more devices are heterogenous devices ([0065] wherein at least a smart phone and a cloud server are described, heterogenous devices).
As per claim 6, Uchiyama discloses the method of claim 1, wherein the one or more heuristics includes minimizing a latency ([0141] wherein the latency (delay in the prior art) is minimized).
As per claim 7, Uchiyama discloses the method of claim 1, wherein the one or more heuristics includes maximizing a throughput ([0141]-[0142] and [Fig. 11] wherein the Splitting mode table describes maximizing throughput in case different situations occur).
As per claim 8, Uchiyama discloses the method of claim 1, wherein each slice of the plurality of slices is associated with only one device ([Fig. 11] wherein each slice is associated with a device (node in the prior art)).
As per claim 9, Uchiyama discloses the method of claim 1, wherein at least two of the slices of the plurality of slices are associated with a first device of the one or more devices ([0065] wherein there is a situation where all lawyers are performed by the communication terminal, meaning all layers are performed by the first device).
As per claim 10, Uchiyama discloses the method of claim 1, wherein the slices are comprised of instructions for transmitting results associated with an execution of slice to a subsequent device ([0117] wherein the slices handler of a range, is instructed on who to transmit the results for subsequent computation). .
As per claim 11, Uchiyama discloses a splitter device comprising: at least one processor ([0290]),and at least one memory coupled to the processor ([0290]), wherein the processor is configured to perform the method of claim 1. Thus, claim 11 is rejected for the same rationale and reasoning as claim 1.
As per claim 12, claim 12 is the device that performs the method of claim 2 and is rejected for the same rationale and reasoning.
As per claim 13, claim 13 is the device that performs the method of claim 3 and is rejected for the same rationale and reasoning.
As per claim 14, claim 14 is the device that performs the method of claim 4 and is rejected for the same rationale and reasoning.
As per claim 15, claim 15 is the device that performs the method of claim 5 and is rejected for the same rationale and reasoning.
As per claim 16, claim 16 is the device that performs the method of claim 6 and is rejected for the same rationale and reasoning.
As per claim 17, claim 17 is the device that performs the method of claim 7 and is rejected for the same rationale and reasoning.
As per claim 18, claim 18 is the device that performs the method of claim 8 and is rejected for the same rationale and reasoning.
As per claim 19, claim 196 is the device that performs the method of claim 9 and is rejected for the same rationale and reasoning.
As per claim 20, claim 20 is the device that performs the method of claim 10 and is rejected for the same rationale and reasoning.
Response to Arguments
Applicant's arguments filed 11 May 2026 have been fully considered but they are not persuasive.
As per claim 1, Applicant states in REMARKS, page 7, that:
“Independent claim 1 recites "generating an intermediate representation of the neural network." Uchiyama does not disclose an intermediate representation. While Uchiyama does disclose a neural network, Uchiyama does not disclose any intermediate representation of the neural network.”
However, Uchiyama discloses intermediate representations by dividing the totality of layers into separate computable layers, wherein the output would represent an intermediate representation of the neural network. The result of a layer would be an “intermediate representation,” representing some portion of the result, but not the totality.
As described in [0079] of Uchiyama and cited by the Examiner, Uchiyama separates a series of computations of the neural network. As described in [0080] of Uchiyama, each layer outputs a certain amount of data, and allows for different delay and also different amounts of data produced for each layer/intermediate representation. Thus, each of layer generates an intermediate representation (output), and Uchiyama teaches the claimed invention.
Applicant’s intentions become clear in REMARKS, page 8, which states “None of these splits are described or otherwise disclosed by Uchiyama to be a representation of the entire neural network as required by the pending claims.” First, the claim language does not recite that the intermediate representation must represent the entire neural network. This actually goes against the common definition of an intermediate representation, which is a representation somewhere in the middle of the entire network. In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., the intermediate representation being representative of the entire neural network) are not recited in the rejected claims. Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993). Here, even the claims describe calculating layers to be divided into groups/slices. Claim 1 recites,
“generating an intermediate representation of a neural network, wherein the neural network is comprised of a plurality of layers…
determining a plurality of slices based on the neural network features, the one or more heuristics and the device characteristics, wherein each slice of the plurality of slices is associated with at least one of the devices, and wherein each of the plurality of slices is associated with one or more of the plurality of layers; and generating the plurality of slices. =“
Generating an intermediate representation of a neural network comprised of a plurality of layers, would be generating the layers individually. The slices are determined by the features (performance in the prior art) of the layers. Since Uchiyama divides the layers and places them into the slices, Uchiyama discloses the claims as recited above.
Applicant is encouraged to clarify what is occurring in the first step to generate the “intermediate representation,” including of what exactly makes up the “intermediate representation,” if it represents the entire neural network to overcome the prior art on record.
Applicant goes on to state in REMARKS, page 8:
“Uchiyama does not disclose or address "extracting ... a plurality of neural network features based on the intermediate representation" nor "determining a plurality of slices based on the neural network features [extracted based on the intermediate representation], ..." nor, then, "generating the plurality of slices."”
Uchiyama discloses "extracting ... a plurality of neural network features based on the intermediate representation" by extracting the delay and output size for each layer (See [0080] of Uchiyama). Examiner has interpreted the intermediate representation to be a layer and thus, the delay and output size would be features of those layers.
Uchiyama discloses "determining a plurality of slices based on the neural network features [extracted based on the intermediate representation], ..." since the slices are determined based on the delay and output size of the intermediate representations (layers in the prior art). (See [0141] of Uchiyama wherein the delay and data load is used to divide the layers and minimize delay).
Finally, the slices are then generated by the grouping of layers using the features (delays and data amounts) as described in [0079] of Uchiyama.
Therefore, Uchiyama properly teaches the limitations above. All other REMARKS regarding claim 1 describe the same arguments above and the same rationale and reasoning applies. Claims 11 is the device that performs the method of claim 1 and the same rationale and reasoning applies.
As per claims 3 and 13, Applicant states for REMARKS, page 11, that Uchiyama. “does not disclose the recited "classifier neural network" of the dependent claim 3 or what is required of the instant claim: "selecting, via a classifier of the splitter device, one or more heuristics based on the neural network features", "wherein the classifier is comprised of a classifier neural network."”
Claims 3 and 13 both describe the classifier in claims 1 and 11 respectively, but here can be interpreted as the invention in its entirety. The selection of heuristics is performed via “a classifier” and “that classifier” comprising a “classifier neural network.” Thus, both only need to be recited in the same invention. In Uchiyama, the heuristics are selected based on the delay and size, and the neural network can be described as a classifier network , classifying the cat as a cat. Thus, the claim is properly rejected.
Again, Examiner is encouraged to clarify what the “intermediate representation” was in relation to the neural network, and most importantly, how the “intermediate representation” is a representation of the entire neural network to overcome the prior art on record and further prosecution. Any limitation providing that the intermediate representation is more than just a layer would further prosecution. If questions remain, please contact the Examiner for an interview.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KANNAN SHANMUGASUNDARAM whose telephone number is (571)270-7763. The examiner can normally be reached M-F 9:00 AM -6:00 PM.
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, Charles Rones can be reached at (571) 272-4085. 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.
/KANNAN SHANMUGASUNDARAM/Primary Examiner, Art Unit 2168
Prosecution Insights