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 .
DETAILED ACTION
Claims 1-48 are present in this application. Claims 20-21 and 23-48 are cancelled. Claims 1-19 and 22 are pending in this office action.
This office action is NON-FINAL.
Drawings
The Drawings filed on 12/20/23 are acceptable for examination purposes.
Specification
The Specification filed on 12/20/23 is acceptable for examination purposes.
Information Disclosure Statement
The information disclosure statements (IDS) filed on 05/22/25, 01/19/24 and 12/20/23 has been considered by the Examiner and made of record in the application file.
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 1-19 and 22 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claim 1 recites “identify one or more straggler devices among a plurality of user devices; suspend transmission of an aggregated model to the one or more straggler devices for local model training; and resume the transmission of the aggregated model to at least one of the one or more straggler devices for the local model training”.
The limitation of “identify one or more straggler devices among a plurality of user devices; suspend transmission of an aggregated model to the one or more straggler devices for local model training; and resume the transmission of the aggregated model to at least one of the one or more straggler devices for the local model training”. That is, other than reciting, “processor,” nothing in the claim element precludes the step from practically being performed in the mind. Accordingly, the claim recites an abstract idea. This judicial exception is not integrated into a practical application. In particular, the claim only recites one additional element – using processor to perform identifying, suspending, and resuming steps. The processor in each steps is recited at a high-level of generality such that it amounts no more than mere instructions to apply the exception using a generic computer component. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The claim is directed to an abstract idea. The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional element of using processor to perform identifying, suspending, and resuming steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer
component cannot provide an inventive concept. The claim is not patent eligible.
Claim 2 is dependent on claim 1 and includes all the limitations of claim 1. Claim
2 recites wherein at least one user device of the plurality of user devices is identified as the one or more straggler devices based on a delay in the apparatus receiving a locally trained machine learning model exceeding a first threshold time period for more than a second threshold number of consecutive iterations in claim 2. But receiving a locally trained machine learning model exceeding a first threshold time period for more than a second threshold number of consecutive iterations does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 3 is dependent on claim 2 and includes all the limitations of claim 2. Claim
3 recites wherein the resuming the transmission comprises transmitting the aggregated model to at least one of the one or more straggler devices…wherein the delay in receiving the locally trained machine learning model exceeds the first threshold time period in claim 3. But transmitting the aggregated model to at least one of the one or more straggler devices does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 4 is dependent on claim 1 and includes all the limitations of claim 1. Claim
4 recites wherein transmit an indication to the identified one or more straggler devices informing that the one or more straggler devices have been identified as a straggler in claim 4. But transmit an indication to the identified one or more straggler devices informing that the one or more straggler devices have been identified does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 5 is dependent on claim 1 and includes all the limitations of claim 1. Claim
5 recites after suspending the transmission of the aggregated model to the one or more straggler devices, evaluate at least one of: a network condition in which the apparatus and the plurality of user devices are operating, or a computational power for the one or more straggler devices to support the local model training of the one or more straggler devices.in claim 5. But after suspending the transmission of the aggregated model to the one or more straggler devices, evaluate at least one of: a network condition in which the apparatus and the plurality of user devices does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 6 is dependent on claim 5 and includes all the limitations of claim 5. Claim
6 recites wherein the resuming the transmission of the aggregated model to the at least one of the one or more straggler devices for the local model training is based on the evaluation indicating that the at least one of the network condition or the computational power to support the local model training is above a third threshold in claim 6. But the resuming the transmission of the aggregated model to the at least one of the one or more straggler devices for the local model training is based on the evaluation does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 7 is dependent on claim 6 and includes all the limitations of claim 6. Claim
7 recites wherein the evaluation indicating that the at least one of the network condition or the computational power is above the third threshold is performed when a federated learning process is not finished while suspending the transmission in claim 7. But the evaluation indicating that the at least one of the network condition or the computational power is above the third threshold is based on the evaluation does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 8 is dependent on claim 5 and includes all the limitations of claim 5. Claim
8 recites transmit a context inquiry message to the one or more straggler devices requesting information on the at least one of the network condition or the computational power to support local model training; and receive a response message from the one or more straggler devices in claim 8. But transmit a context inquiry message to the one or more straggler devices requesting information on the at least one of the network condition or the computational power to support local model training does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 9 is dependent on claim 8 and includes all the limitations of claim 8. Claim
8 recites wherein the response message comprises an acknowledgement or non-acknowledgement message indicating information to determine whether to resume the transmission of the aggregated model to the at least one of the one or more straggler devices in claim 9. But determine whether to resume the transmission of the aggregated model to the at least one of the one or more straggler devices does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 10 is dependent on claim 9 and includes all the limitations of claim 9. Claim 10 recites wherein upon the response message comprising the non-acknowledgement message, a timer is provided from the one or more straggler devices or another network entity to the apparatus which enables the apparatus to transmit the context inquiry message again once the timer expires in claim 10. But transmit the context inquiry message again once the timer expires devices does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 11 is dependent on claim 9 and includes all the limitations of claim 9. Claim 11 recites wherein the at least one memory stores instructions that, when executed by the at least one processor, cause the apparatus to: upon receiving the response message comprising the acknowledgement message, resume the transmission of the aggregated model to the at least one of the one or more straggler devices for the local model training in claim 11. But upon receiving the response message comprising the acknowledgement message, resume the transmission of the aggregated model to the at least one of the one or more straggler devices for the local model training does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 12 is dependent on claim 1 and includes all the limitations of claim 1. Claim 12 recites wherein the one or more straggler devices are user devices which cause a delay in the apparatus receiving a locally trained machine learning model for more than a fourth threshold number of times in a timer window of a defined number of iterations in claim 12. But upon receiving a locally trained machine learning model for more than a fourth threshold number of times in a timer window of a defined number of iterations does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 13 recites “transmit, to a network entity, a locally trained model generated from local model training; receive, from the network entity, an indication that the apparatus is identified as a straggler device and that transmission of an aggregated model is suspended; and resume receiving, from the network entity, the aggregated model based on determining a reduction in a delay in transmitting the locally trained model”.
The limitation of “transmit, to a network entity, a locally trained model generated from local model training; receive, from the network entity, an indication that the apparatus is identified as a straggler device and that transmission of an aggregated model is suspended; and resume receiving, from the network entity, the aggregated model based on determining a reduction in a delay in transmitting the locally trained model”. That is, other than reciting, “processor,” nothing in the claim element precludes the step from practically being performed in the mind. Accordingly, the claim recites an abstract idea. This judicial exception is not integrated into a practical application. In particular, the claim only recites one additional element – using processor to perform transmitting, receiving, and resuming steps. The processor in each steps is recited at a high-level of generality such that it amounts no more than mere instructions to apply the exception using a generic computer component. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The claim is directed to an abstract idea. The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional element of using processor to perform transmitting, receiving, and resuming steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer
component cannot provide an inventive concept. The claim is not patent eligible.
Claim 14 is dependent on claim 13 and includes all the limitations of claim 13. Claim 14 recites wherein the indication that the apparatus is identified as the straggler device is based on the delay in the network entity in receiving the locally trained model exceeding a first threshold time period for more than a second threshold number of consecutive iterations in claim 14. But identified as the straggler device is based on the delay in the network entity in receiving the locally trained model exceeding a first threshold time period for more than a second threshold number of consecutive iterations does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 15 is dependent on claim 13 and includes all the limitations of claim 13. Claim 15 recites collect training data for the locally trained model during a time period between the receiving the indication that transmission of the aggregated model is suspended and the resuming receiving the aggregated model in claim 15. But collect training data for the locally trained model during a time period between the receiving the indication that transmission of the aggregated model is suspended and the resuming receiving the aggregated model does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 16 is dependent on claim 13 and includes all the limitations of claim 13. Claim 16 recites receive a context inquiry message from the network entity requesting information on at least one of: a network condition or a computational power of the apparatus to support local model training in claim 16. But receive a context inquiry message from the network entity requesting information on at least one of: a network condition or a computational power of the apparatus to support local model training does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 17 is dependent on claim 16 and includes all the limitations of claim 16. Claim 17 recites transmit a response message to the network entity comprising the information on the at least one of the network condition or the computational power of the apparatus, wherein the information comprises at least one of a timer or a network condition indicator in claim 17. But transmit a response message to the network entity comprising the information on the at least one of the network condition or the computational power of the apparatus does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 18 is dependent on claim 16 and includes all the limitations of claim 16. Claim 18 recites wherein the response message comprises an acknowledgement or non-acknowledgement message indicating information for the network entity to determine whether to resume transmission of the aggregated model in claim 18. But indicating information for the network entity to determine whether to resume transmission of the aggregated model does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 19 is dependent on claim 18 and includes all the limitations of claim 19. Claim 18 recites upon the response message comprising the non-acknowledgement message, transmit, to the network entity, a timer with the response message; and monitor to receive, from the network entity, the context inquiry message again once the timer expires in claim 19. But upon the response message comprising the non-acknowledgement message, transmit, to the network entity, a timer with the response message; and monitor to receive, from the network entity, the context inquiry message again once the timer expires does not go beyond the abstract idea itself. There are no additional components in the claim that would make it significantly more than the abstract idea.
Claim 22 recites “identifying, by an apparatus, one or more straggler devices among a plurality of user devices; suspending, by the apparatus, transmission of an aggregated model to the one or more straggler devices for local model training; and resuming, by the apparatus, the transmission of the aggregated model to at least one of the one or more straggler devices for the local model training”.
The limitation of “identifying, by an apparatus, one or more straggler devices among a plurality of user devices; suspending, by the apparatus, transmission of an aggregated model to the one or more straggler devices for local model training; and resuming, by the apparatus, the transmission of the aggregated model to at least one of the one or more straggler devices for the local model training”. That is, other than reciting, “apparatus,” nothing in the claim element precludes the step from practically being performed in the mind. Accordingly, the claim recites an abstract idea. This judicial exception is not integrated into a practical application. In particular, the claim only recites one additional element – using apparatus to perform identifying, suspending, and resuming steps. The apparatus in each steps is recited at a high-level of generality such that it amounts no more than mere instructions to apply the exception using a generic computer component. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The claim is directed to an abstract idea. The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional element of using apparatus to perform identifying, suspending, and resuming steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. The claim is not patent eligible.
Claim Rejections 35 U.S.C. §103
6. 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 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.
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 following is a quotation of 35 U.S.C. 103 which forms the basis for all
obviousness rejections set forth in this Office action:
Claims 1-19 and 22 are rejected under 35 U.S.C. 103 as being unpatentable over OKUNO (US 2023/0141483 A1) in view of Qiao et al. (US 2018/0300171 A1).
Regarding claim 1, OKUNO teaches an apparatus, comprising:
at least one processor, (See OKUNO paragraph [0066], processors); and
at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to, (See OKUNO paragraph [0072], The processor 11 loads at least part of the programs in the storage device 13 onto the memory 12 and executes the loaded program):
identify one or more straggler devices among a plurality of user devices, (See OKUNO paragraph [0050], The training job deployment unit 201 manages, for example, processing time by each worker 6 and detects occurrence of a straggler);
suspend transmission of an aggregated model to the one or more straggler devices for local model training, (See OKUNO paragraph [0147], early stopping may be carried out at an early stage when degradation in accuracy is generated by straggler mitigation).
OKUNO does not explicitly disclose resume the transmission of the aggregated model to at least one of the one or more straggler devices for the local model training. However, Qiao teaches resume the transmission of the aggregated model to at least one of the one or more straggler devices for the local model training, (See Qiao paragraph [0041], the ML program computation, (1) the D module suspends execution of micro-tasks from the leaving physical computing unit's Les…and the PS module stops tracking the leaving physical computing unit's LSes, and (5) the D module restarts the suspended micro-tasks).
It would have been obvious to one with ordinary skill in the art before the
effective filing date of the claimed invention was made, to modify resume the transmission of the aggregated model to at least one of the one or more straggler devices for the local model training of Qiao for computing machine learning software applications in a distributed computing environment.
Regarding claim 2, OKUNO taught the apparatus of claim 1, as described above. OKUNO further teaches wherein at least one user device of the plurality of user devices is identified as the one or more straggler devices, (See OKUNO paragraph [0050], The training job deployment unit 201 manages, for example, processing time by each worker 6 and detects occurrence of a straggler); based on a delay in the apparatus receiving a locally trained machine learning model, (See OKUNO paragraph [0042], The training job includes inputting training data to a machine learning model and training of the machine learning model), exceeding a first threshold time period for more than a second threshold number of consecutive iterations, (See OKUNO paragraph [0055], The training job deployment unit 201 determines that a worker 6 whose calculation time τi is greater than a threshold is a straggler).
Regarding claim 3, OKUNO taught the apparatus of claim 2, as described above. OKUNO further teaches the at least one memory stores instructions that, when executed by the at least one processor, cause the apparatus to, (See OKUNO paragraph [0072], The processor 11 loads at least part of the programs in the storage device 13 onto the memory 12 and executes the loaded program):: receive the locally trained machine learning model, (See OKUNO paragraph [0042], The training job includes inputting training data to a machine learning model and training of the machine learning model), from the at least one user device, wherein the delay in receiving the locally trained machine learning model exceeds the first threshold time period, (See OKUNO paragraph [0055], The training job deployment unit 201 determines that a worker 6 whose calculation time τi is greater than a threshold is a straggler).
OKUNO does not explicitly disclose wherein: the resuming the transmission comprises transmitting the aggregated model to at least one of the one or more straggler devices.
However, Qiao teaches wherein: the resuming the transmission comprises transmitting the aggregated model to at least one of the one or more straggler devices, (See Qiao paragraph [0041], the ML program computation, (1) the D module suspends execution of micro-tasks from the leaving physical computing unit's Les…and the PS module stops tracking the leaving physical computing unit's LSes, and (5) the D module restarts the suspended micro-tasks),
It would have been obvious to one with ordinary skill in the art before the
effective filing date of the claimed invention was made, to modify wherein: the resuming the transmission comprises transmitting the aggregated model to at least one of the one or more straggler devices; of Qiao for computing machine learning software applications in a distributed computing environment.
Regarding claim 4, OKUNO taught the apparatus of claim 1, as described above. OKUNO further teaches wherein the at least one memory stores instructions that, when executed by the at least one processor, (See OKUNO paragraph [0072], the programs to be executed by the calculation node 10 may be stored in the storage device 13. The processor 11 loads at least part of the programs in the storage device 13 onto the memory 12 and executes the loaded program), cause the apparatus to: transmit an indication to the identified one or more straggler devices informing that the one or more straggler devices have been identified as a straggler, (See OKUNO paragraph [0204], sign A indicates the embodiment in a case where the leader worker is provided. In the case where the leader worker is provided, each worker 6 transmits information desired for the straggler detection/separation).
Regarding claim 5, OKUNO taught the apparatus of claim 1, as described above. OKUNO further teaches wherein the at least one memory stores instructions that, when executed by the at least one processor, (See OKUNO paragraph [0072], the programs to be executed by the calculation node 10 may be stored in the storage device 13. The processor 11 loads at least part of the programs in the storage device 13 onto the memory 12 and executes the loaded program), cause the apparatus to: after suspending the transmission of the aggregated model to the one or more straggler devices, (See OKUNO paragraph [0147], early stopping may be carried out at an early stage when degradation in accuracy is generated by straggler mitigation); evaluate at least one of: a network condition in which the apparatus and the plurality of user devices are operating, or a computational power for the one or more straggler devices to support the local model training of the one or more straggler devices, (See OKUNO paragraph [0138], the training job deployment unit 201 calculates a threshold that serves as the condition for the straggler. The calculation of the threshold may be realized by using various known techniques. For example, the threshold may be determined based on an average of the calculation time of the workers).
Regarding claim 6, OKUNO taught the apparatus of claim 5, as described above. OKUNO further teaches for the local model training is based on the evaluation indicating that the at least one of the network condition or the computational power to support the local model training is above a third threshold, (See OKUNO paragraph [0138], the training job deployment unit 201 calculates a threshold that serves as the condition for the straggler. The calculation of the threshold may be realized by using various known techniques. For example, the threshold may be determined based on an average of the calculation time of the workers)..
OKUNO does not explicitly disclose wherein the resuming the transmission of the aggregated model to the at least one of the one or more straggler devices.
However, Qiao teaches wherein the resuming the transmission of the aggregated model to the at least one of the one or more straggler devices, (See Qiao paragraph [0041], the ML program computation, (1) the D module suspends execution of micro-tasks from the leaving physical computing unit's Les…and the PS module stops tracking the leaving physical computing unit's LSes, and (5) the D module restarts the suspended micro-tasks),
It would have been obvious to one with ordinary skill in the art before the
effective filing date of the claimed invention was made, to modify wherein the resuming the transmission of the aggregated model to the at least one of the one or more straggler devices of Qiao for computing machine learning software applications in a distributed computing environment.
Regarding claim 7, OKUNO taught the apparatus of claim 6, as described above. OKUNO further teaches wherein the evaluation indicating that the at least one of the network condition or the computational power is above the third threshold, (See OKUNO paragraph [0138], the training job deployment unit 201 calculates a threshold that serves as the condition for the straggler. The calculation of the threshold may be realized by using various known techniques. For example, the threshold may be determined based on an average of the calculation time of the workers).
OKUNO does not explicitly disclose is performed when a federated learning process is not finished while suspending the transmission.
However, Qiao teaches is performed when a federated learning process is not finished while suspending the transmission, (when the system needs to create a recovery checkpoint, the D module will suspend execution of the ML program by waiting for all the LEs to finish their current tasks, and pauses all queued tasks from each LE. The D module takes a checkpoint consisting of: (1) a copy of the D module's state).
It would have been obvious to one with ordinary skill in the art before the
effective filing date of the claimed invention was made, to modify is performed when a federated learning process is not finished while suspending the transmission of Qiao for computing machine learning software applications in a distributed computing environment.
Regarding claim 8, OKUNO taught the apparatus of claim 5, as described above. OKUNO further teaches wherein the at least one memory stores instructions that, when executed by the at least one processor, (See OKUNO paragraph [0072], the programs to be executed by the calculation node 10 may be stored in the storage device 13. The processor 11 loads at least part of the programs in the storage device 13 onto the memory 12 and executes the loaded program), cause the apparatus to: transmit a context inquiry message to the one or more straggler devices requesting information, (See OKUNO paragraph [0204], each worker 6 transmits information desired for the straggler detection/separation and the gradient prediction to the leader worker), on the at least one of the network condition or the computational power to support local model training; and receive a response message from the one or more straggler devices, (See OKUNO paragraph [0138], the training job deployment unit 201 calculates a threshold that serves as the condition for the straggler. The calculation of the threshold may be realized by using various known techniques. For example, the threshold may be determined based on an average of the calculation time of the workers).
Regarding claim 9, OKUNO taught the apparatus of claim 8, as described above.
OKUNO does not explicitly disclose wherein the response message comprises an acknowledgement or non-acknowledgement message indicating information to determine whether to resume the transmission of the aggregated model to the at least one of the one or more straggler devices.
However, Qiao teaches wherein the response message comprises an acknowledgement or non-acknowledgement message indicating information to determine whether to resume the transmission of the aggregated model to the at least one of the one or more straggler devices, (See Qiao paragraph [0041], the ML program computation, (1) the D module suspends execution of micro-tasks from the leaving physical computing unit's Les…and the PS module stops tracking the leaving physical computing unit's LSes, and (5) the D module restarts the suspended micro-tasks).
It would have been obvious to one with ordinary skill in the art before the
effective filing date of the claimed invention was made, to modify wherein the response message comprises an acknowledgement or non-acknowledgement message indicating information to determine whether to resume the transmission of the aggregated model to the at least one of the one or more straggler devices of Qiao for computing machine learning software applications in a distributed computing environment.
Regarding claim 10, OKUNO taught the apparatus of claim 9, as described above. OKUNO further teaches wherein upon the response message comprising the non-acknowledgement message, a timer is provided from the one or more straggler devices, (See OKUNO paragraph [0055], The training job deployment unit 201 determines that a worker 6 whose calculation time τi is greater than a threshold is a straggler), or another network entity to the apparatus which enables the apparatus to transmit the context inquiry message again once the timer expires, (See OKUNO paragraph [0121], the fact that a preset number of epochs has been reached may be set as the end condition, or the fact that a predetermined value of accuracy (training accuracy) of the machine learning model has been reached may end).
Regarding claim 11, OKUNO taught the apparatus of claim 9, as described above. OKUNO further teaches wherein the at least one memory stores instructions that, when executed by the at least one processor, cause the apparatus to, (See OKUNO paragraph [0072], The processor 11 loads at least part of the programs in the storage device 13 onto the memory 12 and executes the loaded program)::
OKUNO does not explicitly disclose upon receiving the response message comprising the acknowledgement message, resume the transmission of the aggregated model to the at least one of the one or more straggler devices for the local model training.
However, Qiao teaches upon receiving the response message comprising the acknowledgement message, resume the transmission of the aggregated model to the at least one of the one or more straggler devices for the local model training, (See Qiao paragraph [0041], the ML program computation, (1) the D module suspends execution of micro-tasks from the leaving physical computing unit's Les…and the PS module stops tracking the leaving physical computing unit's LSes, and (5) the D module restarts the suspended micro-tasks).
It would have been obvious to one with ordinary skill in the art before the
effective filing date of the claimed invention was made, to modify upon receiving the response message comprising the acknowledgement message, resume the transmission of the aggregated model to the at least one of the one or more straggler devices for the local model training of Qiao for computing machine learning software applications in a distributed computing environment.
Regarding claim 12, OKUNO taught the apparatus of claim 9, as described above. OKUNO further teaches wherein the one or more straggler devices are user devices which cause a delay, (See OKUNO paragraph [0010], a synchronization delay may be generated by the straggler, leading to a significantly increase in training time), in the apparatus receiving a locally trained machine learning model, (See OKUNO paragraph [0042], The training job includes inputting training data to a machine learning model and training of the machine learning model). for more than a fourth threshold number of times in a timer window of a defined number of iterations, (See OKUNO paragraph [0055], The training job deployment unit 201 determines that a worker 6 whose calculation time τi is greater than a threshold is a straggler).
Regarding claim 13, OKUNO teaches an apparatus, comprising:
at least one processor, (See OKUNO paragraph [0066], processors); and
at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to, (See OKUNO paragraph [0072], The processor 11 loads at least part of the programs in the storage device 13 onto the memory 12 and executes the loaded program)::
transmit, to a network entity, a locally trained model generated from local model training, (See OKUNO paragraph [0047], the training jobs transmitted by one or more client apparatuses 4 via the network 5 are stored in a queue of the system management unit 2);
receive, from the network entity, an indication that the apparatus, (See OKUNO paragraph [0060], the training job management unit 202 receives an execution result of the training job for the machine learning model from each worker 6 and responds to the client apparatus) is identified as a straggler device and that transmission of an aggregated model is suspended, (See OKUNO paragraph [0147], early stopping may be carried out at an early stage when degradation in accuracy is generated by straggler mitigation); and
OKUNO does not explicitly disclose resume receiving, from the network entity, the aggregated model based on determining a reduction in a delay in transmitting the locally trained model.
However, Qiao teaches resume receiving, from the network entity, the aggregated model based on determining a reduction in a delay in transmitting the locally trained model, (See Qiao paragraph [0041], the ML program computation, (1) the D module suspends execution of micro-tasks from the leaving physical computing unit's Les…and the PS module stops tracking the leaving physical computing unit's LSes, and (5) the D module restarts the suspended micro-tasks).
It would have been obvious to one with ordinary skill in the art before the
effective filing date of the claimed invention was made, to modify resume receiving, from the network entity, the aggregated model based on determining a reduction in a delay in transmitting the locally trained model of Qiao for computing machine learning software applications in a distributed computing environment.
Regarding claim 14, OKUNO taught the apparatus of claim 13, as described above. OKUNO further teaches wherein the indication that the apparatus is identified as the straggler device, (See OKUNO paragraph [0050], The training job deployment unit 201 manages, for example, processing time by each worker 6 and detects occurrence of a straggler); is based on the delay in the network entity in receiving the locally trained model, (See OKUNO paragraph [0042], The training job includes inputting training data to a machine learning model and training of the machine learning model), exceeding a first threshold time period for more than a second threshold number of consecutive iterations, (See OKUNO paragraph [0055], The training job deployment unit 201 determines that a worker 6 whose calculation time τi is greater than a threshold is a straggler). .
Regarding claim 15, OKUNO taught the apparatus of claim 13, as described above. OKUNO further teaches wherein the at least one memory stores instructions that, when executed by the at least one processor, cause the apparatus to, (See OKUNO paragraph [0072], The processor 11 loads at least part of the programs in the storage device 13 onto the memory 12 and executes the loaded program):: collect training data for the locally trained model during a time period, (See OKUNO paragraph [0072], periodically create a recovery checkpoint using a heartbeat mechanism, recover from the last recovery checkpoint upon failure discovery), between the receiving the indication that transmission of the aggregated model is suspended and the resuming receiving the aggregated model, (See Qiao paragraph [0041], the ML program computation, (1) the D module suspends execution of micro-tasks from the leaving physical computing unit's Les…and the PS module stops tracking the leaving physical computing unit's LSes, and (5) the D module restarts the suspended micro-tasks),
Regarding claim 16, OKUNO taught the apparatus of claim 13, as described above. OKUNO further teaches wherein the at least one memory stores instructions, when executed by the at least one processor, , (See OKUNO paragraph [0072], the programs to be executed by the calculation node 10 may be stored in the storage device 13. The processor 11 loads at least part of the programs in the storage device 13 onto the memory 12 and executes the loaded program), cause the apparatus to: receive a context inquiry message from the network entity requesting information on at least one of: a network condition or a computational power of the apparatus to support local model training, (See OKUNO paragraph [0138], the training job deployment unit 201 calculates a threshold that serves as the condition for the straggler. The calculation of the threshold may be realized by using various known techniques. For example, the threshold may be determined based on an average of the calculation time of the workers).
Regarding claim 17, OKUNO taught the apparatus of claim 16, as described above. OKUNO further teaches wherein the at least one memory stores instructions that, when executed by the at least one processor, (See OKUNO paragraph [0072], the programs to be executed by the calculation node 10 may be stored in the storage device 13. The processor 11 loads at least part of the programs in the storage device 13 onto the memory 12 and executes the loaded program), cause the apparatus to: transmit a response message to the network entity comprising the information, (See OKUNO paragraph [0204], each worker 6 transmits information desired for the straggler detection/separation and the gradient prediction to the leader worker), on the at least one of the network condition or the computational power of the apparatus, wherein the information comprises at least one of a timer or a network condition indicator, (See OKUNO paragraph [0138], the training job deployment unit 201 calculates a threshold that serves as the condition for the straggler. The calculation of the threshold may be realized by using various known techniques. For example, the threshold may be determined based on an average of the calculation time of the workers).
Regarding claim 18, OKUNO taught the apparatus of claim 16, as described above.
OKUNO does not explicitly disclose wherein the response message comprises an acknowledgement or non-acknowledgement message indicating information for the network entity to determine whether to resume transmission of the aggregated model.
However, Qiao teaches wherein the response message comprises an acknowledgement or non-acknowledgement message indicating information for the network entity to determine whether to resume transmission of the aggregated model, (See Qiao paragraph [0041], the ML program computation, (1) the D module suspends execution of micro-tasks from the leaving physical computing unit's Les…and the PS module stops tracking the leaving physical computing unit's LSes, and (5) the D module restarts the suspended micro-tasks).
It would have been obvious to one with ordinary skill in the art before the
effective filing date of the claimed invention was made, to modify wherein the response message comprises an acknowledgement or non-acknowledgement message indicating information for the network entity to determine whether to resume transmission of the aggregated model of Qiao for computing machine learning software applications in a distributed computing environment.
Regarding claim 19, OKUNO taught the apparatus of claim 18, as described above. OKUNO further teaches wherein the at least one memory stores instructions that, when executed by the at least one processor, (See OKUNO paragraph [0072], the programs to be executed by the calculation node 10 may be stored in the storage device 13. The processor 11 loads at least part of the programs in the storage device 13 onto the memory 12 and executes the loaded program), cause the apparatus to :upon the response message comprising the non-acknowledgement message, (See OKUNO paragraph [0055], The training job deployment unit 201 determines that a worker 6 whose calculation time τi is greater than a threshold is a straggler), transmit, to the network entity, a timer with the response message; See OKUNO paragraph [0204], each worker 6 transmits information desired for the straggler detection/separation and the gradient prediction to the leader worker), and monitor to receive, from the network entity, (See OKUNO paragraph [0078], A monitor 14a is coupled to the graphic processing device 14), the context inquiry message again once the timer expires, (See OKUNO paragraph [0121], the fact that a preset number of epochs has been reached may be set as the end condition, or the fact that a predetermined value of accuracy (training accuracy) of the machine learning model has been reached may end).
Regarding claim 22. OKUNO teaches a method, comprising:
identifying, by an apparatus, one or more straggler devices among a plurality of user devices, (See OKUNO paragraph [0050], The training job deployment unit 201 manages, for example, processing time by each worker 6 and detects occurrence of a straggler);
suspending, by the apparatus, transmission of an aggregated model to the one or more straggler devices for local model training, (See OKUNO paragraph [0147], early stopping may be carried out at an early stage when degradation in accuracy is generated by straggler mitigation); and
OKUNO does not explicitly disclose resume the transmission of the aggregated model to at least one of the one or more straggler devices for the local model training. However, Qiao teaches resuming, by the apparatus, the transmission of the aggregated model to at least one of the one or more straggler devices for the local model training, (See Qiao paragraph [0041], the ML program computation, (1) the D module suspends execution of micro-tasks from the leaving physical computing unit's Les…and the PS module stops tracking the leaving physical computing unit's LSes, and (5) the D module restarts the suspended micro-tasks).
It would have been obvious to one with ordinary skill in the art before the
effective filing date of the claimed invention was made, to modify resuming, by the apparatus, the transmission of the aggregated model to at least one of the one or more straggler devices for the local model training of Qiao for computing machine learning software applications in a distributed computing environment.
Conclusions/Points of Contacts
The prior art made of record and not relied upon is considered pertinent to
applicant’s disclosure. See form PTO-892.
JAFARKHANI et al. (US 2024/0354589 A1), allows the handling of clients that are slow to respond or have limited communication capabilities. Embodiments also provide operations to address challenges of asynchronous learning such as stale gradients and stragglers, which need to be handled properly to ensure performance.
WANG; et al. (US 2025/0259102 A1) the straggler may be device 1, due to changes in processing and communication latencies between rounds t and t+1. Additionally, as shown in FIG. 2, for non-straggler worker nodes, each of the training rounds may include a waiting period, during which the non-straggler nodes are neither processing nor communicating, but instead are waiting on the straggler to complete its processing and communication activities for that round.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MULUEMEBET GURMU whose telephone number is (571)270-7095. The examiner can normally be reached M-F 9am - 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, Tony Mahmoudi can be reached at 5712724078. 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.
/MULUEMEBET GURMU/Primary Examiner, Art Unit 2163
Prosecution Insights