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 .
Examiner’s Note
In order to expedite prosecution, and to disqualify potential prior art under USC 35 USC 102(b)(2)(C), it would help if the applicant could provide a statement regarding common ownership to the references referred to below. Please see MPEP 717.02(a)(B) where it describes this exception. In particular, please keep in mind that any such statement should be:
Clear and conspicuous (e.g., on a separate paper) to ensure the examiner notices the statement. For example, an attorney or agent of record receives an Office action for Application X in which all the claims are rejected based upon subject matter disclosed in Patent A (either alone or in combination with other references) wherein Patent A is only available as prior art under 35 U.S.C. 102(a)(2). In response to the Office action, the attorney or agent of record for Application X states, in a clear and conspicuous manner, that:[AltContent: rect]
"Application X and Patent A were, not later than the effective filing date of the claimed invention in Application X, owned by Company Z."
Potential Prior Art:
Adrian et al. (2025/0077864)
Ngo et al. (2025/0033195)
Domokos et al. (2025/0036967)
Sun et al. (2024/0037416)
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.
Claims 1-3, 5-6 and 8-10 are rejected under 35 U.S.C. 103 as being unpatentable over Bischoff et al. (2021/0122038) in view of Sivakumar et al. (2021/0287084) and Chi et al. (2024/0046107).
With regard to claim 1, Bischoff teaches a method for adapting a machine learning model to a changed control situation, comprising:
detecting sensor data elements in the changed control situation (Pars. 8, 10, 17, 29, 34, 65-67 - Bischoff teaches that an initial state is determined via a sensor, and trains a machine learning routine that will lead to a target state (i.e., the state change would describe a changed control situation);
However, Bischoff doesn’t teach, for each sensor data element of the detected sensor data elements:
generating multiple augmentations of the sensor data element,
generating, for each augmentation, a respective output by means of a first instance of the machine learning model,
ascertaining a loss between an output of a second instance for the sensor data element and the ascertained target output; and
adapting a second instance of the machine learning model in order to reduce a total loss, which contains the ascertained losses.
Sivakumar teaches of a system for training a machine learning model, where multiple augmentations are generated, and where each augmentation may include a process (output) that changes one or more aspects of each data instance within a training data set (Abstract, pars. 64, 67). Sivakumar further teaches that a first output of the model trained with an augmented sample set is determined, and a second output of another instance of the model trained with a non-augmented sample set is determined (par. 72). The process can then determine a validation loss for the trained model (Fig. 9, pars. 73-74). The process can then determine a weight assigned to a predetermined augmentation for the training data set, and further, assign a greater weight (adapt) if a first improvement in accuracy is greater than a second improvement in accuracy based on first and second augmentations (Fig. 9, pars. 74, 101-102).
Bischoff and Sivakumar don’t specifically teach ascertaining a target output for the sensor data element by combining the generated outputs.
Chi teaches of a system that obtains training samples to query a plurality of AI models, combines the outputs of those models, and adapts a target AI model via knowledge distillation using the combined outputs (Abstract). Chi further teaches that it can evaluate the loss of a target model and update (adapt) a plurality of parameters based on the loss to minimize the loss (pars. 12-13).
It would have been obvious to a person skilled in the art at the time the invention was made to include the teachings of Sivakumar and Chi into the system taught by Bischoff. This would have been obvious because Bischoff, Sivakumar and Chi all teach of training machine learning routines/models, where Bischoff (par. 3) and Chi (par. 4) specifically teach about using those routines in robots, and further, Sivakumar teaches that by generating and applying augmentations, it will provide better quality training data (pars. 2-3). Chi teaches that by combining the outputs produced by the AI models allows for knowledge distillation, which allows knowledge from AI models to be transferred to other models for ease of implementation (par. 5).
With regard to claim 2, Sivakumar teaches the method according to claim 1, further comprising adapting the first instance of the machine learning model toward the adapted second instance of the machine learning model (Pars. 64, 67, 72-74 – A weight is assigned to a predetermined augmentation for the training data set, based on the desired performance. This can change one or more aspects of each data instance within the training data set. Each output based on an instance can be assigned a greater weight based on accuracy of the compared outputs).
With regard to claim 3, Bischoff, Sivakumar and Chi teach the method in claim 1, which includes a lot of the limitations in claim 3. Further, Chi teaches that the learning process also includes batches (pars. 102-104). Chi also teaches that in order to smooth the meta gradient and stabilize the training, a batch of episodes are processed before each meta-update (par. 107). A person skilled in the art would have understood this to mean that the respective instances of each batch processed would be the instance of the subsequent batch in the sequence.
With regard to claim 5, Bischoff teaches the method according to claim 1, wherein the sensor data elements are image data elements (Pars. 29, 34, 80 – The states are determined by a sensor, and the detection of an object (or surrounding area) via a camera may be a target state. This teaches that the sensor data element can be an image data element).
With regard to claim 6, Bischoff teaches the method according to claim 5, wherein the change in the control situation to which the adaptation is made is a change of a camera and/or a change of one or more conditions of an image capture using a camera with which the image data elements are captured (Pars. 29, 34, 80 – The initial state is determined by a sensor, and the detection of an object (or surrounding area) via a camera may be a target state. The camera detects a specified region of the surrounding area of the robot. The detected data of the surrounding area and the alignment of the robot are provided as input data. The states may be continuously updated by the sensor data ).
With regard to claims 8, 9 and 10, Bischoff, Sivakumar and Chi teach the method in claims 1 and 2 above, which includes similar limitations in claim 8. Further, Bischoff teaches that the method controls a robotic device (Fig. 3, pars. 2, 3, 29, 80, 81). Bischoff also teaches that the machine learning routine may be carried out for more than one target state and/or initial state, and a combination of action steps (pars. 22-23). This would teach detecting one or more further sensor data elements in the control situation; processing the one or more further sensor data elements using the adapted second instance of the machine learning model or a first instance of the machine learning model that has been adapted toward the adapted second instance.
Allowable Subject Matter
Claims 4 and 7 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. See PTO-892.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SCOTT T BADERMAN whose telephone number is (571) 272-3644. The examiner can normally be reached 6:00AM -3:00PM, M-Th, every other Friday off.
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, John Cottingham, can be reached at 571-272-1400. 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.
/SCOTT T BADERMAN/Supervisory Patent Examiner, Art Unit 2118
Prosecution Insights