DETAILED ACTION
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 .
Claims 1, 3-4, 9, 13-14, 16 and 18-19 have been amended. Claims 1-20 have been examined.
Response to Arguments
The rejection of claims 1, 9 and 16 under 35 USC § 112 has been withdrawn in view of the claim amendments.
Applicant's arguments, see p. 7 filed 3/10/2026 have been fully considered but they are not persuasive. Applicant argues with respect to the rejection under 35 USC § 112 that the variables found in the claims are found in the specification and would be understood by one of ordinary skill in the art. Applicant’s argument is conclusory without any further reasoning or support. Variables, by their nature, can represent a large range of data. Without a definition, one cannot be certain as to what they might represent in a particular context. The rejection of claims maintained.
Applicant’s arguments, see pp. 7-8, filed 3/10/2026, with respect to the rejection(s) of independent claims 1, 9 and 16 under 35 USC § 103, have been fully considered and are persuasive. These arguments apply equally to all dependent claims. Therefore, the rejections have been withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of U.S. Patent Application Publication 20190130279 by Beggel et al.
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.
Claims 2-5, 7-8, 11-15 and 17-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claims 2, 5, 7, 11-12, 15, 17 and 20 contain one or more of the following mathematical terms and symbols:
θ
,
n
,
x
, a,
L
n
θ
,
L
a
θ
,
train, and test. While their use in the context of the claims in view of the specification is understood as being directed to a variety of machine learning variables and symbols, none of these terms have been specifically defined and therefore it is not clear what they must represent.
Claims 3-5, 8, 13-14, and 18-19 are each rejected as carrying the limitations of a rejected base claim.
Claim Rejections - 35 USC § 103
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.
Claim(s) 1-3, 5, 9, 12-13 and 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication 20230089481 by Liu et al. ("Liu") in view of U.S. Patent 8321220 to Chotimongkol et al. ("Chotimongkol") and U.S. Patent Application Publication 20190130279 by Beggel et al. ("Beggel").
In regard to claim 1, Liu discloses:
1. A method of training a control system comprising: See at least Fig. 6A, depicting a training method.
receiving a data set of N samples that includes normal and unlabeled unidentified anomalous data samples, wherein N is a total number of samples in the data set; Liu, ¶ 0021, “For an attributed network, the set of labeled abnormal nodes is denoted as VL and the set of unlabeled nodes is represented as VU. Note that V={VL, VU} and in the present problem |VL|<<|VU| since only few-shot labeled data is given.” Fig. 2A element 10 and ¶ 0024, “an arbitrary individual network (e.g., an input network 10) with limited labeled data.” Also ¶ 0072, “the plurality of target nodes includes a plurality of labeled anomalous nodes and unlabeled anomalous nodes.”
processing, via a model, the data set to produce an anomaly score associated with each sample in the data set; Liu, Fig. 2A, elements 110, 120, and 220 and ¶ 0029, “In some embodiments, the abnormality valuator module 120 can be built with two feed-forward layers that transform the low-dimensional latent node representations 210 to a set of scalar anomaly scores 220: …” Also ¶ 0072, “assign a scalar anomaly score to the node.”
ranking the normal and anomalous data samples according to the anomaly score associated with each data sample to produce a ranked order; labeling a fraction α of the N samples that have the highest scores with an anomaly label and the remaining samples with a normal label, ¶0023, “Ideally, anomalies that are detected should have higher ranking scores than that of the normal nodes” Also ¶ 0063, “Precision@K is defined as the proportion of true anomalies in a ranked list of K objects. The present system obtains the ranking list in descending order according to the anomaly scores that are computed from a specific anomaly detection algorithm.”
Liu does not expressly disclose:
wherein α is a hyper parameter with a predefined proportion between 0 and 1; This is taught by Beggel. See ¶ 0043, e.g. “The anomaly decision threshold can be chosen based on reconstruction error and can depend on the distribution of reconstruction errors obtained during training. Several alternatives are possible to determine this threshold, several examples of which include: … 3. an adaptive threshold that depends on the expected anomaly rate α, such as the (1−α)% percentile.” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use Beggel’s anomaly rate with Liu’s training in order to provide anomaly determination that aligns with an expected anomaly rate according to user preference as suggested by Beggel.
Liu and Beggel also teach:
retraining the model using all N samples, the labels, and a joint loss function; Liu, Fig. 2A, element 130 and Fig. 2B elements 30 and 100, depicting multiple training rounds. Also ¶ 0033, “Then, the deviation loss module 130 evaluates a final objective function derived from contrastive loss by replacing a distance function with the deviation in Eq. (6): …” ¶ 0037, “Specifically, Meta-GDN 300 extracts meta-knowledge of ground-truth anomalies from different few-shot network anomaly detection tasks on the plurality of auxiliary networks 30 during the training phase, and can be further fine-tuned for the new task on the target network (e.g., the input network 10), such that the anomaly detection system 100 can make fast and effective adaptation.” Also see Beggel, ¶ 0040, “In step 33a, as in step 23, a One-class Support Vector Machine with regularization parameter β computed from the update-step anomaly rate v is trained, but all images previously labeled (especially by user feedback) as normal or anomalies are excluded.”
repeating the processing, ranking, labeling, and retraining steps until the ranked order and labels for all of the N samples … [are within a threshold] ; and Liu, Fig. 2B, depicting repeated training of auxiliary networks. Also see ¶ 0034, “By minimizing the above loss function, the GDN of the anomaly detection system 100 will push the anomaly scores of normal nodes of the input network 10 as close as possible to μr, while enforcing a large positive deviation of at least m between μr and the anomaly scores of abnormal nodes of the input network 10.”
Liu does not expressly disclose: retraining until parameters … do not change. This is taught by Chotimongkol. See Chotimongkol col. 6, lines 45-47, “Using this model, the very same training data can be automatically re-labeled and the model can be retrained, until the training set labels converge.” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use Chotimongkol’s training convergence with Liu’s training in order to provide model training with minimal human intervention as suggested by Chotimongkol (see col. 3, lines 7-11).
Liu also discloses:
outputting the trained model. Liu, Fig. 2B, element 10, depicting a trained network model.
In regard to claim 2, Liu also discloses:
2. The method of claim 1, wherein the joint loss function is expressed by
L
θ
,
y
=
∑
i
=
1
N
1
-
y
i
L
n
θ
x
i
+
y
i
L
a
θ
x
i
. See Liu ‘481, ¶ 0033, “
Then, the deviation loss module 130 evaluates a final objective function derived from contrastive loss by replacing a distance function with the deviation in Eq. (6):
L
=
1
-
y
i
∙
d
e
v
v
i
+
y
i
∙
m
a
x
(
0
,
m
-
d
e
v
(
v
i
)
)
where yi is the ground-truth label of input node vi of the input network 10. If node vi of the input network 10 is an abnormal node, yi=1, otherwise, yi=0. Note that m is a confidence margin which defines a radius around the deviation.”
Also see ¶ 0039 which describes the “meta-objective function” as a sum of the joint loss functions.
In regard to claim 3, Liu also discloses:
3. The method of claim 2, wherein yi = 0 for a normal label and yi = 1 for an anomaly label, wherein yi is a classification label assigned to an i-th data sample, the label indicated whether the sample is normal or anomalous. Liu, ¶ 0033, “If node vi of the input network 10 is an abnormal node, yi=1, otherwise, yi=0.”
In regard to claim 5, Liu also discloses:
5. The method of claim 2, wherein the anomaly scores is expressed by
S
i
t
r
a
i
n
=
L
n
θ
x
i
-
L
a
θ
x
i
. See Liu, section IV(C) on p. 7, right column, “Based on the above property, for each node vi, we can simply define the anomaly score as the difference value between its negative and positive score: f(vi)=si(−)−si(+) .“
In regard to claim 9, Liu discloses:
9. A device control system comprising: a controller configured to, See Liu, Fig. 7, depicting a control system including a controller 500.
…
operate the device control system based on the trained model. See Liu, ¶ 0073, “FIG. 7 is a schematic block diagram of an example device 500 that may be used with one or more embodiments described herein, e.g., as a component of the system and/or as a computing device.”
All further limitations of claim 9 have been addressed in the above rejection of claim 1.
In regard to claims 12-13 and 15 parent claim 9 is addressed above.
All further limitations of claims 12-13 and 15 have been addressed in the above rejections of claims 2-3 and 5, respectively.
Claim(s) 4 and 14 is/are rejected under 35 U.S.C. 103 as being unpatentable over Liu in view of Chotimongkol and Beggel as applied above, and further in view of U.S. Patent Application Publication 20190120719 by Koizumi et al. ("Koizumi").
In regard to claim 4, Liu also discloses:
4. The method of claim 2, wherein yi = 0 for a normal label and … wherein yi is a classification label assigned to an i-th data sample, the label indicated whether the sample is normal or anomalous. Liu, ¶ 0033, “If node vi of the input network 10 is an abnormal node, yi=1, otherwise, yi=0.”
Liu does not expressly disclose: yi = 0.5 for an anomaly label. However, this is taught by Koizumi. See ¶ 0043, “That is, when σ{g(A,x-ϕ)} is equal to or greater than 0.5, it is determined to be anomalous; when it is equal to or smaller than 0.5, it is determined to be normal.” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use Koizumi’s label with Liu’s loss score in order to utilize a simple calculation when determining anomalous data as suggested by Koizumi (see ¶ 0042).
In regard to claim 14, parent claim 12 is addressed above.
All further limitations of claim 14 have been addressed in the above rejection of claim 4.
Claim(s) 6-8, 10-11, 16-18 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Liu in view of Chotimongkol and Beggel as applied above, and further in view of U.S. Patent Application Publication 20200233060 by Lull et al. ("Lull").
In regard to claim 6, Liu does not expressly disclose:
6. The method of claim 1, wherein the data set is time series data received from a sensor that is an optical sensor, an automotive sensor, or an acoustic sensor. This is taught by Lull. See Lull, Fig. 2, depicting automotive sensors 34A-34D. Also Fig. 3, depicting time series data from a sensor. Also see ¶ 0020, “The vehicles 20A and 20B may be equipped with sensor data anomaly detection systems 22A and 22B, respectively. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use Lull’s sensor data with Liu’s anomaly detection in order to provide corrected data for safe vehicle control as suggested by Lull (see ¶ 0018 and ¶ 0034).
In regard to claim 7, Liu also discloses:
7. The method of claim 6 further including … [a control system ] based on the trained model wherein the anomaly scores during operation is expressed by
S
i
t
e
s
t
=
L
n
θ
x
i
. See Liu, ¶ 0033, e.g. “With the reference score μr, the deviation between an anomaly score of node vi of the input network 10 and the reference score can be defined by the deviation loss module 130 in the form of standard score: dev(v-i) = si - μr / σr ” Note that a broad interpretation allows Liu to teach the claimed limitations.
Liu does not expressly disclose: controlling a vehicle. This is taught by Lull. See Lull, ¶ 0018, “vehicle control system.” Also ¶ 0034, “For example, the control system 40 will be provided with a corrected signal to not utilize signals that contain anomalies that may cause an inappropriate movement of the vehicle 20. This inappropriate movement of the vehicle could result in a safety issue to the occupants of the vehicle 20 or others near the vehicle 20.” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use Liu’s anomaly detection and Lull’s vehicle control in order to maintain safe operation as suggested by Lull.
In regard to claim 8, Liu also discloses:
8. The method of claim 7, wherein the fraction α is based on … a parameter being sensed and the unlabeled anomalous data samples. Liu teaches sensed parameters in terms of network types. See ¶ 0004, “For instance, in a citation network that represents citation relations between papers, there are some research papers with a few spurious references (i.e., edges). In a social network that represents friendship of users, there may exist camouflaged users who randomly follow different users, rendering properties like homophily not applicable to this type of relationships.” Also 0063, “Precision@K is defined as the proportion of true anomalies in a ranked list of K objects. The present system obtains the ranking list in descending order according to the anomaly scores that are computed from a specific anomaly detection algorithm.”
Liu does not expressly disclose: … the sensor and … This is taught by Lull. See Lull ¶ 0004, “Environmental factors may also contribute to sensor data anomalies such as thermal inversions, electromagnetic pulse (“EMP”) or electromagnetic radiation (“EMR”), shadowing, contaminated sensor such as water, dirt or smudge on a camera or LIDAR lens, ice build up on radar, can all produce abnormalities in a sensor signal.” Also see ¶ 0032-0033, “The anomaly detection module 54 may determine that the embedded signal 70 is an anomaly and should not be part of the signal 60 based on a comparison to signals 62 and 64, which do not include the embedded signal 70. Once this analysis is complete, the anomaly detection module 54 may then confirm that the potential anomalies 66, 68, and 70 are actual anomalies.” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use Lull’s teaching of sensor specific anomalies with Liu’s proportion anomaly calculation in order to maintain safety as suggested by Lull (see ¶ 0034).
In regard to claim 10, parent claim 9 is addressed above.
All further limitations of claim 10 have been addressed in the above rejection of claim 6.
In regard to claim 11, Liu does not expressly disclose:
11. The device control system of claim 10, wherein the device is a vehicle and the system controls acceleration and deceleration of the vehicle based on the trained model … However, this is taught by Lull. See ¶ 0027, “The actuator 52A may be a throttle actuator that controls the forward and/or rearward movement of the vehicle. The actuator 52B may be a braking actuator that applies one or more brakes of the vehicle 20.” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use Liu’s anomaly detection and Lull’s vehicle control in order to maintain safe operation as suggested by Lull.
All further limitations of claim XXX have been addressed in the above rejection of claim 7.
wherein the anomaly scores during operation is expressed by
S
i
t
e
s
t
=
L
n
θ
x
i
.
In regard to claim 16, Liu discloses:
16. A system for performing at least one perception task associated with …, the system comprising: a processor; and a memory including instructions that, when executed by the processor, cause the processor to: See Liu, Fig. 7, depicting a system including memory and a processor.
Liu does not expressly disclose: autonomous control of a vehicle … operate the vehicle based on the trained model. However, this is taught by Lull. See Lull, ¶ 0018, “vehicle control system.” Also ¶ 0034, “For example, the control system 40 will be provided with a corrected signal to not utilize signals that contain anomalies that may cause an inappropriate movement of the vehicle 20. This inappropriate movement of the vehicle could result in a safety issue to the occupants of the vehicle 20 or others near the vehicle 20.” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use Liu’s anomaly detection and Lull’s vehicle control in order to maintain safe operation as suggested by Lull.
All further limitations of claim 16 have been addressed in the above rejection of claim 1.
In regard to claims 17-18 and 20, parent claim 16 is addressed above.
All further limitations of claims 17-18 and 20 have been addressed in the above rejections of claims 2-3 and 5, respectively.
Claim(s) 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Liu in view of Chotimongkol, Beggel and Lull as applied above, and further in view of Koizumi.
In regard to claim 19, parent claim 17 is addressed above.
All further limitations of claim 17 have been addressed in the above rejection of claim 4.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 James D Rutten whose telephone number is (571)272-3703. The examiner can normally be reached M-F 9:00-5:30 ET.
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, Li B Zhen can be reached at (571)272-3768. 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.
/James D. Rutten/Primary Examiner, Art Unit 2121
Prosecution Insights