Philippe BaumardDownload PDFPatent Trials and Appeals BoardApr 14, 20212019005569 (P.T.A.B. Apr. 14, 2021) Copy Citation UNITED STATES PATENT AND TRADEMARK OFFICE UNITED STATES DEPARTMENT OF COMMERCE United States Patent and Trademark Office Address: COMMISSIONER FOR PATENTS P.O. Box 1450 Alexandria, Virginia 22313-1450 www.uspto.gov APPLICATION NO. FILING DATE FIRST NAMED INVENTOR ATTORNEY DOCKET NO. CONFIRMATION NO. 14/660,519 03/17/2015 Philippe Baumard BAU.0003US 1085 21906 7590 04/14/2021 TROP, PRUNER & HU, P.C. PO Box 41790 HOUSTON, TX 77241 EXAMINER COUGHLAN, PETER D ART UNIT PAPER NUMBER 2121 NOTIFICATION DATE DELIVERY MODE 04/14/2021 ELECTRONIC Please find below and/or attached an Office communication concerning this application or proceeding. The time period for reply, if any, is set in the attached communication. Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the following e-mail address(es): tphpto@tphm.com PTOL-90A (Rev. 04/07) UNITED STATES PATENT AND TRADEMARK OFFICE BEFORE THE PATENT TRIAL AND APPEAL BOARD Ex parte PHILIPPE BAUMARD Appeal 2019-005569 Application 14/660,519 Technology Center 2100 Before MAHSHID D. SAADAT, ALLEN R. MacDONALD, and NABEEL U. KHAN, Administrative Patent Judges. SAADAT, Administrative Patent Judge. DECISION ON APPEAL STATEMENT OF THE CASE Pursuant to 35 U.S.C. § 134(a), Appellant1 appeals from the Examiner’s decision to reject claims 1–10 and 12–21. Claim 11 was cancelled. We have jurisdiction under 35 U.S.C. § 6(b). We affirm. 1 We use the word Appellant to refer to “applicant” as defined in 37 C.F.R. § 1.42(a). Appellant identifies the real party in interest as Philippe Baumard, the inventor of the present application. Appeal Br. 1. Appeal 2019-005569 Application 14/660,519 2 CLAIMED SUBJECT MATTER Appellant’s invention relates to the automated generation of autonomous situational awareness and capability to detect incongruous behaviors in a machine and/or interacting machines and within the behavior of their components. Spec. ¶ 15. Representative claim 1 under appeal read as follows; 1. A method comprising: monitoring, by a system including a processor and sensors, behavioral characteristics of at least a first machine component; generating, by the system through unsupervised learning, unsupervised Bayesian behavioral models at respective different stages, the Bayesian behavioral models representing interrelations between events and machine-to-machine interactions between at least the first machine component and at least a further machine component, the Bayesian behavioral models at the respective different stages changed from one another based on discovery of learned variables representing new interactions between machine components and discovery of new machine components; predicting, by the system using the monitored behavioral characteristics and a Bayesian behavioral model of the generated Bayesian behavioral models, an incongruity of a behavior of at least the first machine component and the machine-to-machine interactions, wherein the incongruity is predicted based on determining a singularity of the events and the machine-to- machine interactions, and a discordance between a calculated expectation of the system and the behavior and the machine-to- machine interactions, the discordance assessing a dynamic distortion and distance between the calculated expectation and the behavior and the machine-to-machine interactions, and wherein the predicting is performed without using a previously built normative rule of behavior and machine-to-machine interactions; and performing, by the system, a counter-measure action comprising controlling at least the first machine component to Appeal 2019-005569 Application 14/660,519 3 address a threat with respect to at least the first machine component in response to the predicted incongruity. REFERENCES AND REJECTIONS2 Claims 1, 4, 6, 8, 17, and 18 stand rejected under 35 U.S.C. § 103 as being unpatentable over Duffield et al. (US 2010/0153316 A1; pub. June 17, 2010) (“Duffield”), Kapoor et al. (US 2012/0240185 A1; pub. Sept. 20, 2012) (“Kapoor”), and Miltonberger (US 2010/0094767 A1; pub. Apr. 15, 2010).3 See Final Act. 14–32. Claims 2, 3, 5, 7, 14, and 19 stand rejected under 35 U.S.C. § 103 as being unpatentable over Duffield, Kapoor, Miltonberger, and Chirashnya et al. (US 2002/0019870 A1; pub. Feb. 14, 2002) (“Chirashnya”). See Final Act. 32–39. Claims 9 and 20 stand rejected under 35 U.S.C. § 103 as being unpatentable over Duffield, Kapoor, Miltonberger, and Girouard et al. (US 2005/0278178 A1; pub. Dec. 15, 2005) (“Girouard”). See Final Act. 39–41. Claims 12 and 13 stand rejected under 35 U.S.C. § 103 as being unpatentable over Duffield, Kapoor, Miltonberger, and Kraemer (US 2013/0111547 A1; pub. May 2, 2013). See Final Act. 42–45. Claim 15 stands rejected under 35 U.S.C. § 103 as being unpatentable over Duffield, Kapoor, Miltonberger, and Weidl et al. (US 2005/0043922 A1; pub. Feb. 24, 2005) (“Weidl”). See Final Act. 45– 46. 2 The Examiner has withdrawn the rejection of claims 1–10 and 12–21 under 35 U.S.C. § 101. See Ans. 5. 3 The statement of the rejection omits claim 8, but the body of the rejection addresses claim 8. See Final Act. 20. Appeal 2019-005569 Application 14/660,519 4 Claims 10, 16, and 21 stand rejected under 35 U.S.C. § 103 as being unpatentable over Duffield, Kapoor, Kraemer, Miltonberger, and Gilbert et al. (US 7,530,105 B2; iss. May 5, 2009) (“Gilbert”). See Final Act. 46–54. ANALYSIS We have reviewed the Examiner’s rejections under 35 U.S.C. § 103 in light of Appellant’s contentions and the evidence of record. We agree with the Examiner and highlight the following for emphasis. Independent Claims 1, 8, and 17 The Examiner found Kapoor discloses, inter alia, “the Bayesian behavioral models at the respective different stages changed from one another based on discovery of learned variables representing new interactions between machine components and discovery of new machine components,” as recited in independent claim 1. See Final Act. 16–17 (citing Kapoor ¶¶ 168, 556). As found by the Examiner, Kapoor teaches performing real-time updates to parameters of a self-organizing map of artificial neurons in order to recognize threats to a computer network, where the real-time updating includes adjusting weight vectors of the artificial neurons, and detecting an anomalous data flow from a new or unknown source. See Ans. 10–11 (citing Kapoor ¶¶ 556, 590). Appellant contends Kapoor describes a flow processing facility that uses a set of artificial neurons for pattern recognition, such as a self- organizing map, where pattern recognition relates to stationary learning, i.e., based on use of stationary models. See Appeal Br. 10 (citing Kapoor ¶¶ 14, 27, 46). In contrast, according to Appellant, the claimed “Bayesian behavioral models” of claim 1 are non-stationary models that are “changed from one another based on discovery of learned variables representing new Appeal 2019-005569 Application 14/660,519 5 interactions between machine components and discovery of new machine components,” where learned variables are new variables. See Appeal Br. 10; see also Reply Br. 8–9. In support of its argument, Appellant further cites a Rule 132 Declaration by the inventor Philippe Baumard submitted during prosecution of the present application. See Appeal Br. 7, 10; see also Declaration Under 37 C.F.R. § 1.132, submitted August 16, 2018 (“Rule 132 Declaration”). We are not persuaded by this argument because it is not commensurate with the scope of claim 1. Neither claim 1, nor Appellant’s Specification, limits the claimed “Bayesian behavioral models” to non- stationary models. Similarly, neither claim 1, nor Appellant’s Specification, limits the claimed “learned variables representing new interactions between machine components and discovery of new machine components” to new variables. As the Examiner correctly found, Kapoor teaches performing real-time updates to a self-organizing map of artificial neurons in order to recognize threats to a computer network, where the real-time updating including adjusting weight vectors of artificial neurons and detecting an anomalous data flow from a new or unknown source. See Kapoor ¶¶ 14, 27, 46, 168, 556. Because Appellant does not persuasively distinguish the claimed “Bayesian behavior models” of claim 1 from Kapoor’s self- organizing map, we agree with the Examiner that Kapoor’s adjustment of weight vectors and detection of an anomalous data flow teaches or suggests “the Bayesian behavioral models at the respective different stages changed from one another based on discovery of learned variables representing new interactions between machine components and discovery of new machine components,” as recited in claim 1. Appeal 2019-005569 Application 14/660,519 6 Furthermore, Appellant’s Rule 132 Declaration appears to proffer arguments regarding Kapoor that are similar to Appellant’s arguments described above. See Rule 132 Decl. ¶¶ 3–5. For the reasons previously described, we agree with the Examiner that the arguments proffered in the Rule 132 Declaration are not persuasive of Examiner error. See Ans. 7. The Examiner further found Miltonberger discloses, inter alia, “a discordance between a calculated expectation of the system and the behavior and the machine-to-machine interactions, the discordance assessing a dynamic distortion and distance between the calculated expectation and the behavior and the machine-to-machine interactions, and wherein the predicting is performed without using a previously built normative rule of behavior and machine-to-machine interactions,” as recited in claim 1. See Final Act. 18–19 (citing Miltonberger ¶ 85). As found by the Examiner, Miltonberger teaches determining and representing thresholds corresponding to component risk scores of a parameter of an event (e.g., login), where a probability of a parameter is specific for a particular user based on data specific to that particular user, as opposed to a previously built normative rule of behavior. See Ans. 31–32 (citing Miltonberger ¶ 85). Appellant next contends Miltonberger also refers to use of stationary models and would not have led to use of a non-stationary model as recited in claim 1. See Appeal Br. 16. More specifically, according to Appellant, Miltonberger describes a structure of a predictive user model (“PUM”) as pre-formulated so that there is no requirement to discover the structure of the model, but rather the requirement is to estimate unknown parameters of the model. See Appeal Br. 16–17 (citing Miltonberger ¶ 59); see also Reply Br. 22–23. Appellant further argues, because Miltonberger discloses the structure of the PUM as pre-formulated, Miltonberger teaches the set of Appeal 2019-005569 Application 14/660,519 7 parameters of the PUM is stationary. See Appeal Br. 17. In support of its argument, Appellant further cites its Rule 132 Declaration. See id. This argument is not persuasive either because it is also not commensurate with the scope of claim 1. Similar to its argument regarding Kapoor, Appellant’s argument focuses on Miltonberger’s alleged use of a stationary model as opposed to a non-stationary model, but neither claim 1, nor Appellant’s Specification, limits the claimed “Bayesian behavioral models” to non-stationary models. As the Examiner correctly found, Miltonberger describes a system that determines and represents thresholds corresponding to component risk scores of a parameter of an event (e.g., user login), where the probability of the parameter is specific for a particular user based on data specific to that particular user. See Miltonberger ¶ 85. We agree with the Examiner that Miltonberger’s disclosure of the probability of a parameter of an event teaches or suggests the claimed “discordance between a calculated expectation of the system and the behavior and the machine-to-machine interactions.” See Final Act. 18–19. We further agree that because Miltonberger’s parameter is specific for a particular user based on data specific to that particular user, Miltonberger further teaches or suggests that the calculation of the probability is performed “without using a previously built normative rule of behavior and machine-to-machine interactions.” See Ans. 31–32. Appellant’s argument does not persuasively explain how Miltonberger’s disclosure is distinct from the relevant limitation of claim 1. Further, Appellant’s Rule 132 Declaration appears to proffer arguments regarding Miltonberger that are similar to Appellant’s arguments described above. See Rule 132 Decl. ¶ 6. For the reasons previously Appeal 2019-005569 Application 14/660,519 8 described, we agree with the Examiner that the arguments proffered in the Rule 132 Declaration are not persuasive of Examiner error. See Ans. 7. Thus, we are not persuaded that the Examiner erred in finding the combination of Duffield, Kapoor, and Miltonberger teaches or suggests all the elements of claim 1. No separate arguments are presented for independent claims 8 and 17. See Appeal Br. 17. Thus, we are also not persuaded that the Examiner erred in finding the combination of Duffield, Kapoor, and Miltonberger teaches or suggests all the elements of claims 8 and 17. Dependent Claim 19 The Examiner found the combination of Miltonberger and Chirashnya discloses, inter alia, “predicting performed without using a previously built normative rule that is pre-established by a human, an application, or a machine,” as recited in claim 19. See Final Act. 18, 38–39 (citing Miltonberger ¶ 85; Chirashnya ¶ 7). As found by the Examiner, Miltonberger teaches “predicting performed without using a previously built normative rule” as previously described above. See Final Act. 18–19 (citing Miltonberger ¶ 85); see also Ans. 31–32 (citing Miltonberger ¶ 85). As further found by the Examiner, Chirashnya discloses diagnostic methods for large complex switched networks utilizing Bayesian networks, where one or more probabilities of a malfunction of a given module are compared with a predetermined threshold, where the predetermined threshold teaches “a previously built normative rule that is pre-established by . . . an application.” See Final Act. 38–39 (citing Chirashnya ¶ 7); see also Ans. 32 (citing Chirashnya ¶¶ 27, 47, 63). Appellant contends Chirashnya refers to Bayesian network models that declare a fault in response to a measure exceeding a certain threshold. Appeal 2019-005569 Application 14/660,519 9 See Appeal Br. 18 (citing Chirashnya ¶¶ 10, 27, 47, 63). According to Appellant, such thresholds are examples of a previously built normative rule of behavior that is pre-established by a human, an application, or a machine, and thus, would not have led one of ordinary skill in the art to the claimed limitation, “predicting performed without using a previously built normative rule that is pre-established by a human, an application, or a machine,” as recited in claim 19. See Appeal Br. 18; see also Reply Br. 25. Appellant’s argument is not persuasive because it solely addresses Chirashnya rather than the combination of Miltonberger and Chirashnya. As described above, the Examiner relied upon Miltonberger for teaching “predicting performed without using a previously built normative rule,” and further relied upon Chirashnya for teaching “a previously built normative rule that is pre-established by a human, an application, or a machine.” Appellant’s argument does not address the Examiner’s combination of Miltonberger and Chirashnya, and thus, is not persuasive. One cannot show non-obviousness by attacking references individually when the rejection is based on a combination of references. See In re Merck & Co., Inc., 800 F.2d 1091, 1097 (Fed. Cir. 1986); see also In re Keller, 642 F.2d 413, 425 (CCPA 1981). Thus, we are not persuaded that the Examiner erred in finding the combination of Duffield, Kapoor, Miltonberger, and Chirashnya teaches or suggests all the elements of claim 19. Dependent Claim 10 The Examiner found Gilbert discloses, inter alia, learn over time a persistence of incongruities and incongruous behaviors of the machine components, to detect contrived incongruities and deliberate acclimatization to planned incongruous behavior in the system, the deliberate acclimatization to the planned incongruous behavior in the Appeal 2019-005569 Application 14/660,519 10 system achieved by an entity introducing small variations of incongruous behavior into the system such that the system would not be able to detect the planned incongruous behavior, as recited in claim 10. See Final Act. 47–48 (citing Gilbert at Abstract). As found by the Examiner, Gilbert teaches a platform that utilizes inexact matching to identify, detect, or predict “attack variants” or variants of previous attack strategies based on incomplete data. See Ans. 33 (citing Gilbert 11:18–50). Appellant contends Gilbert merely describes that tactical attacks can be detected using a search of an input graph. See Appeal Br. 21–22 (citing Gilbert at Abstract). Appellant specifically argues that there is nothing in Gilbert that provides any teaching or suggestion of any deliberate acclimatization to a planned incongruous behavior by introducing small variations of incongruous behavior into a system such that the system would not be able to detect the planned incongruous behavior. See Appeal Br. 22; see also Reply Br. 27–28. We are not persuaded by this argument either. As the Examiner correctly found, Gilbert discloses a platform that utilizes inexact matching to identify, detect, or predict “attack variants” (i.e., variants of previous attack strategies based on incomplete data). See Gilbert 11:18–50. We agree with the Examiner that Gilbert’s disclosure teaches or suggests “learn over time a persistence of incongruities and incongruous behaviors of the machine components.” See Ans. 33–34. With respect to, the claimed “deliberate acclimatization to planned incongruous behavior in the system . . . achieved by an entity introducing small variations of incongruous behavior into the system such that the system would not be able to detect the planned incongruous behavior,” Appellant’s argument does not persuasively explain Appeal 2019-005569 Application 14/660,519 11 how Gilbert’s disclosure is distinct from the aforementioned limitation of claim 10. Thus, we are not persuaded that the Examiner erred in finding the combination of Duffield, Kapoor, Kraemer, Miltonberger, and Gilbert teaches or suggests all the elements of claim 10. Remaining Dependent Claims No separate arguments are presented for dependent claims 2–7, 9, 12– 16, 18, 20, and 21. See Appeal Br. 19–20, 22. Accordingly, we sustain the rejections of those claims as well. CONCLUSION We affirm the Examiner’s decision to reject claims 1–10 and 12–21 under 35 U.S.C. § 103. DECISION SUMMARY In summary: Claim(s) Rejected 35 U.S.C. § Reference(s)/Basis Affirmed Reversed 1, 4, 6, 8, 17, 18 103 Duffield, Kapoor, Miltonberger 1, 4, 6, 8, 17, 18 2, 3, 5, 7, 14, 19 103 Duffield, Kapoor, Miltonberger, Chirashnya 2, 3, 5, 7, 14, 19 9, 20 103 Duffield, Kapoor, Miltonberger, Girouard 9, 20 12, 13 103 Duffield, Kapoor, Miltonberger, Kraemer 12, 13 15 103 Duffield, Kapoor, Miltonberger, Weidl 15 10, 16, 21 103 Duffield, Kapoor, Kraemer, 10, 16, 21 Appeal 2019-005569 Application 14/660,519 12 Miltonberger, Gilbert Overall Outcome 1–10, 12–21 TIME PERIOD FOR RESPONSE No time period for taking any subsequent action in connection with this appeal may be extended under 37 C.F.R. § 1.136(a). See 37 C.F.R. § 1.136(a)(1)(iv). AFFIRMED Copy with citationCopy as parenthetical citation