Anoop Deoras et al.Download PDFPatent Trials and Appeals BoardDec 3, 201914016186 - (D) (P.T.A.B. Dec. 3, 2019) 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. 14/016,186 09/02/2013 Anoop Deoras 339419.01 8641 69316 7590 12/03/2019 MICROSOFT CORPORATION ONE MICROSOFT WAY REDMOND, WA 98052 EXAMINER BLANKENAGEL, BRYAN S ART UNIT PAPER NUMBER 2658 NOTIFICATION DATE DELIVERY MODE 12/03/2019 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): chriochs@microsoft.com usdocket@microsoft.com PTOL-90A (Rev. 04/07) UNITED STATES PATENT AND TRADEMARK OFFICE BEFORE THE PATENT TRIAL AND APPEAL BOARD Ex parte ANOOP DEORAS, KAISHENG YAO, XIAODONG HE, LI DENG, GEOFFREY GERSON ZWEIG, RUHI SARIKAYA, DONG YU, MEI-YUH HWANG, and GREGOIRE MESNIL Appeal 2018-004902 Application 14/016,186 Technology Center 2600 Before MICHAEL J. STRAUSS, JEREMY J. CURCURI, and MICHAEL M. BARRY, Administrative Patent Judges. CURCURI, 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, 4, 5, 7–11, and 13–24. 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 Microsoft Technology Licensing, LLC. Appeal Br. 2. Appeal 2018-004902 Application 14/016,186 2 CLAIMED SUBJECT MATTER The claims are directed to “the assignment of semantic labels to words in a natural language utterance.” Spec. ¶ 4. Claim 1, reproduced below, is illustrative of the claimed subject matter: 1. A method comprising a plurality of acts performed by one or more processors of a computing system, the method comprising: receiving a sequence of words from a computer- implemented automatic speech recognition system that is configured to recognize words in a spoken utterance, the sequence of words comprising the words in the spoken utterance; responsive to receiving the sequence of words, retrieving respective pluralities of feature values for words in the sequence of words; providing a plurality of feature values corresponding to a word in the sequence of words as input to a computer- implemented recurrent neural network (RNN), the computer- implemented RNN trained to assign labels to words, the labels comprising at least one semantic label; receiving output of the RNN, the output indicative of a semantic label that is to be assigned to the word, the semantic label identifying a semantic meaning of the word; and assigning the semantic label to the word based upon the output of the RNN. REFERENCES The prior art relied upon by the Examiner is: Appeal 2018-004902 Application 14/016,186 3 Name Reference Date Garfield Garfield et al., “Call classification using recurrent neural networks, support vector machines and finite state automata,” Knowl Inf Syst (2006) 9(2):131–156. July 18, 2005 Melcher US 2010/0217684 A1 Aug. 26, 2010 Normandin US 2007/0219792 A1 Sept. 20, 2007 Collobert US 2011/0301942 A1 Dec. 8, 2011 REJECTIONS Claims 1, 4, 5, 7–11, and 13–24 are rejected under 35 U.S.C. § 101 as directed to a judicial exception without significantly more. Final Act. 6–9. Claims 1, 8, 10, 11, 13, 14, and 19–24 are rejected under 35 U.S.C. § 102(a)(1) as anticipated Garfield. Final Act. 9–16. Claims 4, 9, 15, and 18 are rejected under 35 U.S.C. § 103 as obvious over Garfield and Melcher. Final Act. 16–19. Claims 5, 16, and 17 are rejected under 35 U.S.C. § 103 as obvious over Garfield, Melcher, and Normandin. Final Act. 19–22. Claim 7 is rejected under 35 U.S.C. § 103 as obvious over Garfield and Collobert. Final Act. 22–23. We review the appealed rejections for error based upon the issues identified by Appellant, and in light of the arguments and evidence produced thereon. Ex parte Frye, 94 USPQ2d 1072, 1075 (BPAI 2010) (precedential). Appeal 2018-004902 Application 14/016,186 4 OPINION The Judicial Exception Rejection of Claims 1, 4, 5, 7–11, and 13–24 The Examiner determines claims 1, 4, 5, 7–11, and 13–24 are directed to a judicial exception without significantly more. Final Act. 6–9. The Examiner determines Regarding claims 1, 11, and 20, the RNN is essentially a series of mathematical calculations using formulas described in the specification, such as equations 2-4 on page 14 and equations 8–10 on page 17, making it an abstract idea. Assigning a semantic label is similarly considered a mathematical process, as it involves altering the numerical representation of the object. Receiving words from a speech recognizer is simple data transfer, without any significant transformation taking place. Similarly, retrieving the feature values is mere data transfer. Providing values as input and receiving as output from the RNN is also data transfer, without any significant transformation to the form of the data occurring, merely mathematical processing. The use of processors, memory, or storage is considered well- understood, routine, and conventional in the art, and is not significantly more. Final Act. 7. An invention is patent-eligible if it claims a “new and useful process, machine, manufacture, or composition of matter.” 35 U.S.C. § 101. However, the U.S. Supreme Court has long interpreted 35 U.S.C. § 101 to include implicit exceptions: “[l]aws of nature, natural phenomena, and abstract ideas” are not patentable. Alice Corp. v. CLS Bank Int’l, 573 U.S. 208, 216 (2014) (citation omitted). In determining whether a claim falls within an excluded category, we are guided by the Court’s two-step framework, described in Mayo and Alice. Id. at 217–18 (citing Mayo Collaborative Servs. v. Prometheus Labs., Inc., 566 U.S. 66, 75–77 (2012)). In accordance with that framework, we first Appeal 2018-004902 Application 14/016,186 5 determine what concept the claim is “directed to.” See Alice, 573 U.S. at 219 (“On their face, the claims before us are drawn to the concept of intermediated settlement, i.e., the use of a third party to mitigate settlement risk.”); see also Bilski v. Kappos, 561 U.S. 593, 611 (2010) (“Claims 1 and 4 in petitioners’ application explain the basic concept of hedging, or protecting against risk.”). Concepts determined to be abstract ideas, and, thus, patent ineligible, include certain methods of organizing human activity, such as fundamental economic practices (Alice, 573 U.S. at 219–20; Bilski, 561 U.S. at 611); mathematical formulas (Parker v. Flook, 437 U.S. 584, 594–95 (1978)); and mental processes (Gottschalk v. Benson, 409 U.S. 63, 67 (1972)). Concepts determined to be patent eligible include physical and chemical processes, such as “molding rubber products” (Diamond v. Diehr, 450 U.S. 175, 191 (1981)); “tanning, dyeing, making water-proof cloth, vulcanizing India rubber, smelting ores” (id. at 182 n.7 (quoting Corning v. Burden, 56 U.S. 252, 267–68 (1853))); and manufacturing flour (Benson, 409 U.S. at 69 (citing Cochrane v. Deener, 94 U.S. 780, 785 (1876))). In Diehr, the claim at issue recited a mathematical formula, but the Court held that “a claim drawn to subject matter otherwise statutory does not become nonstatutory simply because it uses a mathematical formula.” Diehr, 450 U.S. at 187; see also id. at 191 (“We view respondents’ claims as nothing more than a process for molding rubber products and not as an attempt to patent a mathematical formula.”). Having said that, the Court also indicated that a claim “seeking patent protection for that formula in the abstract . . . is not accorded the protection of our patent laws, . . . and this principle cannot be circumvented by attempting to limit the use of the formula to a particular technological environment.” Id. (citing Benson and Appeal 2018-004902 Application 14/016,186 6 Flook); see, e.g., id. at 187 (“It is now commonplace that an application of a law of nature or mathematical formula to a known structure or process may well be deserving of patent protection.”). If the claim is “directed to” an abstract idea, we turn to the second step of the Alice and Mayo framework, where “we must examine the elements of the claim to determine whether it contains an ‘inventive concept’ sufficient to ‘transform’ the claimed abstract idea into a patent- eligible application.” Alice, 573 U.S. at 221 (citation omitted). “A claim that recites an abstract idea must include ‘additional features’ to ensure ‘that the [claim] is more than a drafting effort designed to monopolize the [abstract idea].’” Id. (quoting Mayo, 566 U.S. at 77). “[M]erely requir[ing] generic computer implementation[] fail[s] to transform that abstract idea into a patent-eligible invention.” Id. The U.S. Patent and Trademark Office (“USPTO”) recently published revised guidance on the application of § 101. USPTO’s 2019 Revised Patent Subject Matter Eligibility Guidance, 84 Fed. Reg. 50 (Jan. 7, 2019); USPTO October 2019 Update: Subject Matter Eligibility (Oct. 17, 2019). Under that guidance, we first look to whether the claim recites: (1) any judicial exceptions, including certain groupings of abstract ideas (i.e., mathematical concepts, certain methods of organizing human activity such as a fundamental economic practice, or mental processes); and (2) additional elements that integrate the judicial exception into a practical application (see MPEP § 2106.05(a)–(c), (e)–(h) (9th ed. 2018)). See 84 Fed. Reg. at 52–55. Only if a claim (1) recites a judicial exception and (2) does not integrate that exception into a practical application, do we then look to whether the claim: Appeal 2018-004902 Application 14/016,186 7 (3) adds a specific limitation beyond the judicial exception that is not “well-understood, routine, conventional” in the field (see MPEP § 2106.05(d)); or (4) simply appends well-understood, routine, conventional activities previously known to the industry, specified at a high level of generality, to the judicial exception. See 84 Fed. Reg. at 56. Are the claims patent-eligible? Step One Claim 1 is a method claim having several steps, which falls within the “process” category of 35 U.S.C. § 101. Claim 11 is a computer device claim, which falls within the “machine” category of 35 U.S.C. § 101. Claim 20 is a computer-readable storage medium claim, which falls within the “manufacture” category of 35 U.S.C. § 101. Thus, we must determine whether the claims recite a judicial exception, and fail to integrate the exception into a practical application. See 84 Fed. Reg. at 54–55. If both elements are satisfied, the claim is directed to a judicial exception under the first step of the Alice/Mayo test. See id. Step 2A—Prong One Independent claim 1 recites [i] receiving a sequence of words from a computer- implemented automatic speech recognition system that is configured to recognize words in a spoken utterance, the sequence of words comprising the words in the spoken utterance; [ii] responsive to receiving the sequence of words, retrieving respective pluralities of feature values for words in the sequence of words. These steps cover receiving input data and feature extraction, and are thus one or more of “observation, evaluation, judgment, opinion,” which is an example of “concepts performed in the human mind.” 84 Fed. Reg. at 52. Appeal 2018-004902 Application 14/016,186 8 Independent claim 1 further recites “[iii] providing a plurality of feature values corresponding to a word in the sequence of words as input to a computer-implemented recurrent neural network (RNN), the computer- implemented RNN trained to assign labels to words, the labels comprising at least one semantic label.” This step covers receiving input data (at the “RNN”), and is thus an “observation,” which is an example of “concepts performed in the human mind.” 84 Fed. Reg. at 52. Further, the “RNN” describes “mathematical relationships, mathematical formulas or equations, mathematical calculations,” which are examples of “[m]athematical concepts.” 84 Fed. Reg. at 52; see also Spec. 14 (equations 2–4), 17 (equations 8–10). Independent claim 1 further recites “[iv] receiving output of the RNN, the output indicative of a semantic label that is to be assigned to the word, the semantic label identifying a semantic meaning of the word; and [v] assigning the semantic label to the word based upon the output of the RNN.” These steps cover receiving output data, and are thus an “observation,” which is an example of “concepts performed in the human mind.” 84 Fed. Reg. at 52. Put another way, the overall claim describes the process of applying labels (semantic labels) to data (words). The claimed process is the same process that a human performs when understanding spoken language, and is thus one or more of “observation, evaluation, judgment, opinion,” which is an example of “concepts performed in the human mind.” 84 Fed. Reg. at 52. The recited “RNN” is “trained to assign labels to words” but the “RNN” sets forth “mathematical relationships, mathematical formulas or equations, mathematical calculations,” which are examples of “[m]athematical Appeal 2018-004902 Application 14/016,186 9 concepts.” 84 Fed. Reg. at 52; see also Spec. 14 (equations 2–4), 17 (equations 8–10). Step 2A, Prong Two Because claims 1, 11, and 20 recite a judicial exception, we next determine if the claims recite additional elements that integrate the judicial exception into a practical application. In addition to the limitations of claim 1 discussed above that recite abstract concepts, claim 1 further recites “one or more processors of a computing system,” “a computer-implemented automatic speech recognition system,” and “a computer-implemented recurrent neural network (RNN).” The Specification does not provide additional details that would distinguish the additional limitations from a generic implementation. See Spec. ¶¶ 26 (describing automatic speech recognition (ASR) system 114), 30–32 (describing RNN 128), 78–79 (describing processor 1302). We do not find the recited computer-related limitations are sufficient to integrate the judicial exception into a practical application. Specifically, there is no improvement to the functioning of the “one or more processors of a computing system,” “computer-implemented automatic speech recognition system,” and “computer-implemented recurrent neural network (RNN)”, but, instead, devices merely implement the abstract idea. In this case, we do not see any particular machine or manufacture that is integral to the claim; nor do we see any transformation. That is, we do not see any of the additionally recited elements applying or using the judicial exception in any meaningful way beyond generally linking the judicial exception to the recited elements. Accordingly, we determine that claim 1 is directed to a judicial exception because it does not recite additional elements that integrate the recited judicial exception into a practical application. Claims 11 and 20 Appeal 2018-004902 Application 14/016,186 10 recite similar limitations to those discussed with respect to claim 1, and therefore also are directed to a judicial exception. Is there something else in the claims that ensures that they are directed to significantly more than a patent ineligible concept? Step 2B Because claims 1, 11, and 20 are directed to a judicial exception, we next determine, according to Alice, whether these claims recite an element, or combination of elements, that is enough to ensure that the claim is directed to significantly more than a judicial exception. The Specification does not provide additional details about the “one or more processors of a computing system,” “computer-implemented automatic speech recognition system,” and “computer-implemented recurrent neural network (RNN),” that would distinguish the recited hardware components from generic implementation. See Spec. ¶¶ 26 (describing automatic speech recognition (ASR) system 114), 30–32 (describing RNN 128), 78–79 (describing processor 1302). The conventional or generalized functional terms by which the computer components are described reasonably indicate that Appellant’s Specification discloses conventional components. See Spec. ¶¶ 26 (describing automatic speech recognition (ASR) system 114), 30–32 (describing RNN 128), 78–79 (describing processor 1302). In view of Appellant’s Specification, the claimed hardware components, including the “one or more processors of a computing system,” “computer-implemented automatic speech recognition system,” and “computer-implemented recurrent neural network (RNN)” are reasonably determined to be generic, purely conventional computer elements. Thus, the Appeal 2018-004902 Application 14/016,186 11 claims do no more than require generic computer elements to perform generic computer functions, rather than improve computer capabilities. Accordingly, we determine that claims 1, 11, and 20 are not directed to significantly more than a patent ineligible concept. Appellant’s principal arguments Appellant argues the claims “are directed towards assigning a semantic label to a word based upon output of an RNN, and not generally towards ‘converting one form of numerical representation to another’[] as alleged by the Examiner.” Appeal Br. 8–9. Appellant further argues the claims, particularly by the use of an RNN to account for additional context and improve generalization, improve computer capabilities. See Appeal Br. 9–16 (citing Enfish, McRO, and Trading Technologies); see also Reply Br. 2–7. These arguments are not persuasive of Examiner error because, as we explain above, the argued elements of claims 1, 11, and 20 are the abstract concepts of “concepts performed in the human mind” and “[m]athematical concepts.” 84 Fed. Reg. at 52. We also explain that the claims do not recite additional elements that integrate the recited judicial exception into a practical application, and are not directed to significantly more than a patent ineligible concept. Appellant’s identified improvements such as use of the RNN are improvements to the abstract idea because the RNN itself is a mathematical concept, not an improvement to a technology or computer functionality. Thus, on the record before us, the claim limitations do not improve the functionality of the various hardware components, nor do they achieve an improved technological result in conventional industry practice. McRO, Inc. v. Bandai Namco Games Am., Inc., 837 F.3d 1299, 1316 (Fed. Appeal 2018-004902 Application 14/016,186 12 Cir. 2016). More particularly, we do not agree that use of the RNN improves the technology as a whole. See Ans. 17–20. Appellant argues the claims are not directed to an abstract idea when compared to claims previously considered by the Supreme Court and Federal Circuit. Appeal Br. 16–21 (citing Benson, Alice, Enfish, McRO, AmDocs, Trading Technologies, DDR Holdings). These arguments also do not show any error because, as we explain above, the elements of claims 1, 11, and 20 are the abstract concepts of “concepts performed in the human mind” and “[m]athematical concepts.” 84 Fed. Reg. at 52. We also explain that the claims do not recite additional elements that integrate the recited judicial exception into a practical application, and are not directed to significantly more than a patent ineligible concept. We have analyzed Appellant’s claims in light of current USPTO guidance, and the guidance considers claims previously considered by the Supreme Court and Federal Circuit. 84 Fed. Reg. 50 (Jan. 7, 2019). Thus, we do not agree that use of the RNN improves the technology as a whole. See Ans. 21–22. Appellant further argues using the RNN to assign semantic labels to words is a practical application. See Appeal Br. 21–22. Appellant further argues the claims are rooted in technology: Specifically, the subject claims recite features that provide a solution to a problem associated with conventional SLU [(spoken language understanding)] systems - relatively poor generalization over complex patterns of words. This is a problem that specifically arises in the realm of computer-implemented SLU systems (there is no historic real-world analog). To overcome this problem, the claims recite features relating to use of a computer-implemented RNN to assign semantic labels to words. It is thus apparent that the solution is rooted in computer technology. Appeal 2018-004902 Application 14/016,186 13 Appeal Br. 22–23. These arguments do not show any error because we determine the RNN is part of the abstract idea. That is, the “RNN” sets forth “mathematical relationships, mathematical formulas or equations, mathematical calculations,” which are examples of “[m]athmatical concepts.” 84 Fed. Reg. at 52; see also Spec. 14 (equations 2–4), 17 (equations 8–10). That is, the RNN itself is a mathematical concept, not improvements to a technology or computer functionality or a practical application of assigning semantic labels to words. See Ans. 22. Appellant further argues the claims recite steps that are not well- understood, routine, or conventional because the claims recite “using output of an RNN to assign a semantic label to a word.” Appeal Br. 23. This argument does not show any error because, as we explain above, Appellant’s identified improvements such as use of the RNN are improvements to the abstract idea because the RNN itself is a mathematical concept, not improvements to a technology or computer functionality. Further, as we explain above, the claimed hardware components, including the “computer- implemented recurrent neural network (RNN)” are reasonably determined to be generic, purely conventional computer elements. See Spec. ¶¶ 30–32 (describing RNN 128). See Ans. 22. Appellant further argues the following: Applicant has not claimed the alleged abstract concept of “a series of mathematical calculations”, but instead has claimed a specific, new, and useful technique for assigning a semantic label to a word in a sequence of words. Many other techniques for performing mathematical calculations are not preempted by the independent claims. Appeal 2018-004902 Application 14/016,186 14 Appeal Br. 25. This argument does not show any error because the RNN itself is a mathematical concept, and part of the abstract idea. Further, we recognize as a threshold matter that the Supreme Court has described “the concern that drives this exclusionary principle [to statutory patentability] as one of pre-emption.” See Alice, 573 U.S. at 216. Characterizing preemption as a driving concern for patent eligibility, however, is not the same as characterizing preemption as the sole test for patent eligibility. As our reviewing court has explained: “questions on preemption are inherent in and resolved by the [section] 101 analysis,” and, although “preemption may signal patent ineligible subject matter, the absence of complete preemption does not demonstrate patent eligibility.” Ariosa Diagnostics, Inc. v. Sequenom, Inc., 788 F.3d 1371, 1379 (Fed. Cir. 2015); cf. OIP Techs., Inc. v. Amazon.com, Inc., 788 F.3d 1359, 1362–63 (Fed. Cir. 2015) (“[T]hat the claims do not preempt all price optimization or may be limited to price optimization in the e-commerce setting do not make them any less abstract”). See Ans. 23. We, therefore, sustain the Examiner’s rejection under 35 U.S.C. § 101 of claims 1, 4, 5, 7–11, and 13–24. The Anticipation Rejection of Claims 1, 8, 10, 11, 13, 14, and 19–24 by Garfield The Examiner finds Garfield describes all limitations of claim 1. Final Act. 10–11. In particular, the Examiner finds Garfield describes “receiving output of the RNN, the output indicative of a semantic label that is to be assigned to the word” (Final Act. 11 (citing Garfield 142, first paragraph)) and “the semantic label identifying a semantic meaning of the word” as recited in claim 1 (id. (citing Garfield 135, last paragraph)). Appeal 2018-004902 Application 14/016,186 15 Appellant presents the following principal arguments: Regarding “receiving output of the RNN, the output indicative of a semantic label that is to be assigned to the word” as recited in claim 1, Appellant argues [t]he classification of a call to a call class, as output by the neural network of Garfield, cannot fairly be characterized as being indicative of a semantic label that is to be assigned to the word, since each word in the utterance has already been assigned values for each call class. App. Br. 27 (citing Garfield 139, Table 6). Regarding “the output indicative of a semantic label that is to be assigned to the word” as recited in claim 1, Appellant argues “[i]t is unclear how a service to which a caller is to be routed is in any way germane to a label assigned to a word that identifies a semantic meaning of the word as claimed.” App. Br. 27 (citing Garfield 135, last paragraph). In response, the Examiner explains “the training step of Garfield does teach the claimed limitation in the cited portion, where the output of the RNN assigns the call class.” Ans. 23. The Examiner also explains “[t]his semantic label could refer to the call class assigned to the utterance or word, as it is not limited by the claim limitations.” Ans. 23. The Examiner further explains “the semantic meaning of the word is determined by the classification of the input utterance.” Ans. 24. In reply, Appellant argues that the semantic vectors for the words in Garfield “are in no way based upon output of the neural network of Garfield.” Reply Br. 8. Appellant also argues that “[a] service to which a caller is to be routed cannot fairly be characterized as a label assigned to a word that identifies a semantic meaning of the word as claimed.” Reply Br. 8. Appeal 2018-004902 Application 14/016,186 16 Appellant’s arguments persuade us that the Examiner erred in finding Garfield describes “receiving output of the RNN, the output indicative of a semantic label that is to be assigned to the word, the semantic label identifying a semantic meaning of the word” as recited in claim 1. Garfield discloses The input layer has one input unit for each call class. During training and testing, utterances are presented sequentially to the network one word at a time. Each input unit receives the value of υ(w, ci), where ci denotes the particular call class of a word, w, with which the input is associated. Utterances are presented to the network as a sequence of word input and category output representations, one pair for each word. At the beginning of each new utterance, the context layers are cleared and initialised with 0 values. Each unit in the output layer corresponds to a particular call class, and the output unit that represents the desired call class is set to 1 and all other output units are set to 0. Garfield 142, first paragraph; see also Garfield 139, Table 6 (depicting an example simple recurrent neural network input). Based on these disclosures in Garfield, we find the output of the RNN in Garfield does not indicate a semantic label that is to be assigned to the word. Rather, the output of the RNN in Garfield indicates a call class that is assigned to the entirety of the utterance. Further, we find the call class in Garfield does not identify a semantic meaning of the word. Rather, the call class in Garfield indicates a classification of the entirety of the utterance. Thus, the Examiner erred in finding Garfield describes “receiving output of the RNN, the output indicative of a semantic label that is to be assigned to the word, the semantic label identifying a semantic meaning of the word” as recited in claim 1. Appeal 2018-004902 Application 14/016,186 17 We, therefore, do not sustain the Examiner’s anticipation rejection of claim 1. We also do not sustain the Examiner’s anticipation rejection of claims 8, 10, 21, and 23, which depend from claim 1. Independent claim 11 recites “outputting, from the RNN, data that is indicative of a semantic label that is to be assigned to the word, the semantic label output based upon the feature values, the semantic label identifying a semantic meaning of the word.” We, therefore, do not sustain the Examiner’s anticipation rejection of claim 11 for the same reasons discussed above with respect to claim 1. We also do not sustain the Examiner’s anticipation rejection of claims 13, 14, 19, and 24, which depend from claim 11. Independent claim 20 recites “receiving output from the RNN, the output being a probability distribution over a plurality of [semantic] labels assignable to the at least one word.” We, therefore, do not sustain the Examiner’s anticipation rejection of claim 20 for the same reasons discussed above with respect to claim 1. We also do not sustain the Examiner’s anticipation rejection of claim 22, which depends from claim 20. The Remaining Rejections The Examiner’s applications of the remaining references fail to cure the deficiency discussed above with respect to the independent claims. See Final Act. 16–23. We, therefore, do not sustain the Examiner’s obviousness rejections of claims 4, 5, 7, 9, and 15–18. Appeal 2018-004902 Application 14/016,186 18 CONCLUSION The Examiner’s decision to reject claims 1, 4, 5, 7–11, and 13–24 is affirmed. DECISION SUMMARY In summary: Claims Rejected 35 U.S.C. § Reference(s)/Basis Affirmed Reversed 1, 4, 5, 7– 11, 13–24 101 Judicial Exception 1, 4, 5, 7– 11, 13–24 1, 8, 10, 11, 13, 14, 19–24 102 Garfield 1, 8, 10, 11, 13, 14, 19– 24 4, 9, 15, 18 103 Garfield, Melcher 4, 9, 15, 18 5, 16, 17 103 Garfield, Melcher, Normandin 5, 16, 17 7 103 Garfield, Collobert 7 Overall Outcome 1, 4, 5, 7– 11, 13–24 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