Prosecution Insights
Last updated: November 04, 2026
Application No. 18/654,710

SYSTEM, NETWORK AND METHOD FOR SELECTIVE ACTIVATION OF A COMPUTING NETWORK

Non-Final OA §101§103§112
Filed
May 03, 2024
Examiner
LEE, TSU-CHANG
Art Unit
2128
Tech Center
2100 — Computer Architecture & Software
Assignee
Silvretta Research Inc.
OA Round
1 (Non-Final)
73%
Grant Probability
Favorable
1-2
OA Rounds
1y 0m
Est. Remaining
86%
With Interview

Examiner Intelligence

Grants 73% — above average
73%
Career Allowance Rate
316 granted / 434 resolved
+17.8% vs TC avg
Moderate +14% lift
Without
With
+13.6%
Interview Lift
resolved cases with interview
Typical timeline
3y 6m
Avg Prosecution
41 currently pending
Career history
488
Total Applications
across all art units

Statute-Specific Performance

§101
39.1%
-0.9% vs TC avg
§103
31.5%
-8.5% vs TC avg
§102
6.4%
-33.6% vs TC avg
§112
18.5%
-21.5% vs TC avg
Black line = Tech Center average estimate • Based on career data from 434 resolved cases

Office Action

§101 §103 §112
The present application, filed on or after 16 March 2013, is being examined under the first inventor to file provisions of the AIA . DETAILED ACTION This office action is in response to Applicant’s submission filed on 3 May 2024. THIS ACTION IS NON-FINAL. Status of Claims Claims 1-21 are pending. Claims 4-7, 10, 11, 15-16 and 19-21 are rejected under 35 U.S.C. 112(b) as indefinite. Claims 1–12 and 21 are rejected under 35 U.S.C. 101 for being directed to software per se. Claim 13-20 are rejected under 35 U.S.C. 101 for being directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more. Claims 1-5, 8-10, 12-15, 17-20 are rejected under 35 U.S.C. 103 as unpatentable. Claims 17-18 are objected to. There is no art rejection for claims 6-7, 11, 16 and 21. Claim Objections Claims 17 and 18 recite "the method of claim 1." For purposes of this rejection, they are treated as depending from method claim 13. Appropriate correction is required. 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. A claim is indefinite if, when read in light of the specification, it fails to inform, with reasonable certainty, those skilled in the art about the scope of the invention. Nautilus, Inc. v. Biosig Instruments, Inc., 110 USPQ.2d 1688, U.S. Supreme Court (2014). Claims 4-7, 10, 11, 15-16 and 19-21 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 pre-AIA the applicant regards as the invention. Regarding claim 4, Claim 4, (a)the claim recites "cease the sampling of the plurality of computational paths once the output of the stochastic neural network is sufficiently precise." The term "sufficiently precise" is a relative term which renders the claim indefinite. The term is not defined by the claim, and the specification does not provide a standard for ascertaining the requisite degree. The specification states only that sufficient precision "can be addressed using, among various methods, the currently available probability of activation of the output node; the stability of the output values, and in other ways" (¶ [0035]). This leaves the threshold and the measure of precision open-ended. One of ordinary skill in the art would not be reasonably apprised of the scope of the invention. See MPEP 2173.05(b), the claim is therefore indefinite. (b) the claim recites "programming operable by the stochastic neural network." It is unclear how a neural network, which the claim recites as nodes and synapses, operates programming, or what structure executes that programming. The metes and bounds of the claim are therefore unclear, the claim is therefore indefinite. For purposes of examination, the limitation is interpreted as instructions, executed by a processor, that stop sampling additional computational paths when the output satisfies a convergence or accuracy criterion. Regarding claim 5 /15, the claim recites "initiating each of the plurality of computational paths from a first of the at least two hidden nodes, wherein the first of the at least two hidden nodes has been activated by a previous computational path." The limitation is indefinite for two reasons. First, it is unclear whether the "previous computational path" is one of the recited "plurality of computational paths." If it is, the earliest-sampled path of the plurality cannot satisfy the limitation, because no previous path exists to have activated its starting hidden node. The specification indicates that at least one path must first traverse the network from the inputs (¶¶ [0033], [0034]). If it is not, the claim does not set forth where the "previous computational path" comes from or how it relates to the claimed sampling. Second, "a first of the at least two hidden nodes" can be read either as a single hidden node from which every path is initiated, or as a respective hidden node for each path. These readings differ materially in scope. For purposes of examination, the limitation is interpreted as initiating one or more of the computational paths from a hidden node that was activated by an earlier computation. Regarding claims 6, 7 and 16, the claims recite that the selective activation "is scheduled by one or more incoming nodes to occur at a later time which corresponds to a stochastically drawn waiting time of the plurality of computational paths." It is unclear what "a stochastically drawn waiting time of the plurality of computational paths" measures. It could be the time until a particular path is sampled, a single waiting time shared by all of the paths, or something else. The phrase "of the plurality of computational paths" does not make clear whether one waiting time or multiple waiting times are required. For purposes of examination, the limitation is interpreted as an upstream node scheduling the activation of a downstream node at a time drawn from a probability distribution, consistent with ¶¶ [0037]–[0038]. Regarding claims 10 / 19, the claim recite that at least one of the computational paths entering a hidden node "can further activate" one or more nodes positioned closer to the output node. The use of "can" renders the claims indefinite. It is unclear whether the claims require that a path actually activates the downstream nodes, or merely that the path is capable of doing so. See MPEP 2173.05(d) (optional language); MPEP 2173.05(g) (functional limitations). For purposes of examination, the limitation is interpreted as being met by a path that is capable of activating downstream nodes. Regarding claim 11, the claim recites that "each of the plurality of nodes acts independently to balance its probability of activation with the connected nodes." The term "balance" is not defined by the claim, and its meaning in this context is unclear. The claim does not set forth: what quantity is balanced, i.e., against what the node's probability is compared or equalized; what condition constitutes a balanced state; or which "connected nodes" (upstream, downstream, or both) are involved. The specification describes the balancing only as causing a recalculation "if the updated value is sufficiently large" (¶ [0041]). This is itself a term of degree without an objective standard. One of ordinary skill in the art would not be reasonably apprised of the scope of the claim. See MPEP 2173.05(b), the claim is therefore indefinite. Regarding claim 21, the claim recites that the selective activation "is scheduled according to a sampled or expected delay … wherein the sampled or expected delay is stored in a buffer with processing performed on a first-in first-out basis regardless of a computed wait time." It is unclear what is being processed on a first-in, first-out basis: the stored delays, the scheduled activations, or the computational paths. For purposes of examination, the limitation is interpreted as storing delays in a buffer and processing the corresponding activations in first-in, first-out order without waiting for the actual delay to elapse, consistent with ¶¶ [0039] and [0042]. Claims 6 and 10 depend from claim 5. Claims 16, 19 and 20 depend from claim 15. Each inherits the indefiniteness of its base claim and is rejected for the same reasons. 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. Non-Statutory Subject Matter Claims 1–12 and 21 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter. Claim 1 recites "a computing system, comprising: a stochastic neural network comprising: a plurality of nodes; and a plurality of synapses," with a subset of the nodes "selectively activated based on sampling a plurality of computational paths." The claim recites no processor, memory, circuitry, or other physical structure. The specification describes the nodes and synapses in terms of activation weights, routing probabilities, and stochastic functions (¶¶ [0014]–[0022]). It also states that the disclosed aspects may be implemented "entirely software" (¶ [0061]). Under the broadest reasonable interpretation in light of the specification, the claimed "computing system" therefore covers a stochastic neural network embodied solely as software or as a mathematical model. Such a model has no physical or tangible form. It is not a process, machine, manufacture, or composition of matter. See MPEP 2106.03(I); Digitech Image Techs., LLC v. Elecs. for Imaging, Inc., 758 F.3d 1344, 1348–50 (Fed. Cir. 2014); In re Nuijten, 500 F.3d 1346, 1356–57 (Fed. Cir. 2007). Dependent claims 2–12 and 21 add only further functional or mathematical characteristics of the network and supply no physical structure. The "buffer" of claim 21 reasonably covers a software data structure, such as the ordered list of delays described at ¶¶ [0039] and [0042]. It therefore does not cure the deficiency. The examiner suggests amending claim 1 to recite the network as implemented by one or more hardware processors and a non-transitory memory storing instructions. Alternatively, it could recite hardware nodes and synapses containing calculation and sampling circuitry, as described at ¶ [0058]. Judicial Exception Claims 13-20 of the claimed invention are directed to a judicial exception, an abstract idea, without significantly more. Regarding claims 1-20, (Independent Claims) With regards to claim 1, Step 1: The claim is directed to software per se (and it’s dependent claims). For compact prosecution, claims 1-12 and 21 are also analyzed for judiciary exception. Step 2A – Prong 1: the claim, in part, recites: (a) "a subset of the plurality of nodes is selectively activated" (mental process and/or math concept). As drafted, this is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. That is, other than reciting generic computer elements, nothing in the claim element precludes the step from practically being performed in the mind. For example, but for the language about generic computer elements, "selectively activated" encompasses evaluating a graph of nodes and connections and deciding which nodes to mark as active. This is based on observation, evaluation, judgement and/or opinion, and could be performed by a human using paper, pen and calculator, e.g., on a network of the size shown in FIG. 1 (five nodes and six synapses). If a claim limitation, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components, then it falls within the "Mental Processes" grouping of abstract ideas. (b) "based on sampling a plurality of computational paths traversing the stochastic neural network" (math concept). As drafted, this is a process that, under its broadest reasonable interpretation, covers mathematical concepts but for the recitation of generic computer components. Read in light of the specification, "sampling … computational paths" describes mathematical calculations: computing a routing probability for each outgoing synapse, e.g., a SoftMax of a weighted sum of incoming values, or a normalized squared distance from a synapse-specific centroid (¶¶ [0020]–[0022]); and drawing random samples according to those probabilities to determine which nodes are activated (¶¶ [0004], [0019], [0023]). Mathematical calculations and relationships have been found by the courts to be abstract ideas, e.g., see MPEP 2106.04(a)(2)(I). If a claim limitation, under its broadest reasonable interpretation, covers mathematical calculations or relationships, then it falls within the "Mathematical Concepts" grouping of abstract ideas. Accordingly, the claim recites an abstract idea. Step 2A – Prong 2: This judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of: (a) "A computing system", which is mere instructions to implement an abstract idea using a generic computing device, or merely uses a computer as a tool to perform an abstract idea (see MPEP 2106.05(f)); (b) "a stochastic neural network comprising: a plurality of nodes; and a plurality of synapses, wherein each synapse of the plurality of synapses connects a respective pair of the plurality of nodes", which is recited at a high level of generality and amounts to the generic structure of a neural network on which the mathematical operations are performed. This is mere instructions to apply the abstract idea using generic computing structure (see MPEP 2106.05(f)), and generally links the abstract idea to a particular technological environment, i.e., neural networks (see MPEP 2106.05(h)). The specification asserts that selective activation reduces computational load (¶¶ [0002]–[0003], [0048]). However, the claim recites only the result, partial activation, without the particular way that result is achieved. Under its broadest reasonable interpretation, "a subset" includes all of the nodes, so the claim does not require any reduction in computation (see ¶ [0024]). See MPEP 2106.05(a); Two-Way Media Ltd. v. Comcast Cable Commc'ns, LLC, 874 F.3d 1329, 1337 (Fed. Cir. 2017). Accordingly, the additional elements individually or in combination do not integrate the judicial exception into a practical application. The claim is directed to an abstract idea. Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above, the claim recites the additional elements of: (a) "A computing system", which is mere instructions to implement an abstract idea using a generic computing device, or merely uses a computer as a tool to perform an abstract idea (see MPEP 2106.05(f)); (b) "a stochastic neural network comprising: a plurality of nodes; and a plurality of synapses …", which is mere instructions to apply the abstract idea using generic computing structure (see MPEP 2106.05(f)) and field of use and technological environment (see MPEP 2106.05(h)). The specification describes the computing environment only in generic terms (¶¶ [0058], [0061]–[0065]). Accordingly, the additional elements individually or in combination do not amount to significantly more than the judicial exception. The claim is not patent eligible. With regards to claim 13, Step 1: The claim recites a method, which is a process, and falls into one of the statutory categories. Step 2A – Prong 1: the claim, in part, recites: (a) "selectively activating a subset of the plurality of nodes" (mental process and/or math concept). As drafted, this is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. For example, "selectively activating" encompasses evaluating a graph of nodes and connections and deciding which nodes to mark as active. This is based on observation, evaluation, judgement and/or opinion, and could be performed by a human using paper, pen and calculator. It therefore falls within the "Mental Processes" grouping of abstract ideas. (b) "based on sampling a plurality of computational paths traversing the stochastic neural network" (math concept). As drafted, this is a process that, under its broadest reasonable interpretation, covers mathematical concepts but for the recitation of generic computer components. The step of "sampling … computational paths", read in light of the specification (¶¶ [0004], [0019]–[0023]), describes mathematical calculations: computing routing and activation probabilities from weighted sums of incoming values and drawing random samples according to those probabilities. The "plurality of activation weights" are the numerical parameters of those calculations. Mathematical calculations and relationships have been found by the courts to be abstract ideas, e.g., see MPEP 2106.04(a)(2)(I). The limitation therefore falls within the "Mathematical Concepts" grouping of abstract ideas. Accordingly, the claim recites an abstract idea. Step 2A – Prong 2: This judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of: (a) "providing a stochastic neural network comprising a plurality of nodes" and "providing a plurality of synapses, wherein each of the plurality of synapses comprises a connection between a respective pair of the plurality of nodes", which is mere instructions to apply the abstract idea using generic neural-network structure (see MPEP 2106.05(f)) and field of use and technological environment (see MPEP 2106.05(h)); (b) "providing a plurality of activation weights, wherein each of the plurality of activation weights is associated with a respective synapse … or a respective node", which is extra-solution activity of pre-solution data gathering, i.e., setting up the parameters used in the calculation (see MPEP 2106.05(g)). As with claim 1, the claim recites the result of partial activation without the particular way the asserted computational savings are achieved (see MPEP 2106.05(a)). Accordingly, the additional elements individually or in combination do not integrate the judicial exception into a practical application. The claim is directed to an abstract idea. Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above, the claim recites the additional elements of: (a) "providing a stochastic neural network …" and "providing a plurality of synapses …", which is mere instructions to apply the abstract idea using generic neural-network structure (see MPEP 2106.05(f)) and field of use and technological environment (see MPEP 2106.05(h)); (b) "providing a plurality of activation weights …", which is extra-solution activity of pre-solution data gathering (see MPEP 2106.05(g)). The courts have found limitations directed to storing and retrieving information, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), "storing and retrieving information in memory"; Versata Dev. Grp., Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334 (Fed. Cir. 2015)). Accordingly, the additional elements individually or in combination do not amount to significantly more than the judicial exception. The claim is not patent eligible. (Dependent claims) Claims 2-12 and 21 are dependent on claim 1 and include all the limitations of claim 1. Claims 14-20 are dependent on claim 13 and include all the limitations of claim 13. Claims 17 and 18 recite "the method of claim 1"; for purposes of this rejection they are treated as depending from method claim 13. Therefore, claims 2-12 and 14-21 recite the same abstract ideas. With regards to claim 2, the claim recites the further limitation of "each of the plurality of computational paths selectively activates a sequence of nodes, wherein a computed output of each node in the sequence of nodes is dependent upon the probability of such a node being activated for a set of input values" (mental process and/or math concept). This further describes a mathematical relationship between each node's output and its activation probability. It therefore falls within the "Mathematical Concepts" grouping of abstract ideas (see MPEP 2106.04(a)(2)(I)). Except for citing generic computer elements to implement the abstract idea, there is no additional element showing integration into a practical application or adding something significantly more to the abstract idea. The claim is not patent eligible. With regards to claims 3 / 14, the claims recite the further limitation of "an output of the stochastic neural network comprises a sequence of the sampled plurality of computational paths with a corresponding sequence of output values that represent approximations of the output of the stochastic neural network" (mental process and/or math concept. This describes an iterative numerical (Monte Carlo) approximation, in which each additional sample refines an estimated value. It therefore falls within the "Mathematical Concepts" grouping of abstract ideas (see MPEP 2106.04(a)(2)(I)). Except for citing generic computer elements to implement the abstract idea, there is no additional element showing integration into a practical application or adding something significantly more to the abstract idea. The claims are not patent eligible. With regards to claim 4, the claim recites the further limitation of "to …cease the sampling of the plurality of computational paths once the output of the stochastic neural network is sufficiently precise" (mental process and/or math concept. This is a mathematical convergence criterion and an evaluation or judgement of whether a result is sufficiently precise, which could be performed by a human using paper, pen and calculator. It therefore falls within the "Mathematical Concepts" and "Mental Processes" groupings of abstract ideas. The claim recites the additional element of "programming operable by the stochastic neural network", which is merely using a computer as a tool to perform an abstract idea (see MPEP 2106.05(f)). Accordingly, at Step 2A, prong two, the additional elements individually or in combination do not integrate the judicial exception into a practical application. Accordingly, at Step 2A, prong two, the additional elements individually or in combination do not integrate the judicial exception into a practical application. The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above, the additional element of programming operable by the stochastic neural network", which is merely using a computer as a tool to perform an abstract idea (see MPEP 2106.05(f)). Accordingly, at Step 2B the additional elements individually or in combination do not amount to significantly more than the judicial exception. The claim is not patent eligible. With regards to claims 5 / 15, the claims recite the further limitation of "…the plurality of nodes comprises at least one input node, at least one output node and at least two hidden nodes … and wherein sampling the plurality of computational paths comprises initiating each of the plurality of computational paths from a first of the at least two hidden nodes, wherein the first of the at least two hidden nodes has been activated by a previous computational path" (mental process and/or math concept). This further specifies where the mathematical sampling begins, i.e., selecting the starting point of the random draw based on a previously computed activation value (¶¶ [0033]–[0034]). Selecting a starting point in a graph based on a prior result could also be performed by a human using paper, pen and calculator. The limitation therefore falls within the "Mathematical Concepts" and "Mental Processes" groupings of abstract ideas. Except for citing generic computer elements to implement the abstract idea, there is no additional element showing integration into a practical application or adding something significantly more to the abstract idea. The claims are not patent eligible. With regards to claims 6 / 16, the claims recite the further limitation of "the selective activation of a subset of the plurality of nodes is scheduled by one or more incoming nodes to occur at a later time which corresponds to a stochastically drawn waiting time of the plurality of computational paths" (mental process and/or math concept). Read in light of the specification, this describes drawing a random waiting time from a probability distribution, e.g., a geometric distribution with expected value 1/p (¶¶ [0037], [0043]), and ordering the activations by that value. This is a mathematical calculation. Scheduling or ordering events by an assigned time is also an evaluation and judgement that could be performed by a human using paper, pen and calculator. The limitation therefore falls within the "Mathematical Concepts" and "Mental Processes" groupings of abstract ideas. Except for citing generic computer elements to implement the abstract idea, there is no additional element showing integration into a practical application or adding something significantly more to the abstract idea. The claims are not patent eligible. With regards to claim 7, the claim recites the further limitation of " wherein the selective activation of a subset of the plurality of nodes is scheduled by one or more incoming nodes to occur at a later time which corresponds to a stochastically drawn waiting time of the plurality of computational path" (mental process and/or math concept), which is further process of evaluation of data in graph model. Except for citing generic computer elements to implement the abstract idea, there is no additional element showing integration into a practical application or adding something significantly more to the abstract idea. The claim is not patent eligible. With regards to claims 8 / 17, the claims recite the additional element of "the plurality of computational paths is sampled in parallel", which is merely using a computer as a tool to perform an abstract idea, i.e., performing the same mathematical sampling concurrently using generic computing resources (see MPEP 2106.05(f)). Accordingly, at Step 2A, Prong 2, the additional elements individually or in combination do not integrate the judicial exception into a practical application, and the claims are directed to an abstract idea. At Step 2B, for the same reasons, the additional elements individually or in combination do not amount to significantly more than the judicial exception. The claims are not patent eligible. With regards to claims 9 / 18, the claims recite the further limitation of "the plurality of computational paths is sampled independently by each node or each synapse". This further describes the manner of performing the mathematical sampling, i.e., performing the random draws locally at each node or synapse. It therefore falls within the "Mathematical Concepts" grouping of abstract ideas. To the extent the nodes or synapses are regarded as additional elements, they are generic neural-network structure used as a tool to perform the abstract idea (see MPEP 2106.05(f)). Except for citing generic computer elements to implement the abstract idea, there is no additional element showing integration into a practical application or adding something significantly more to the abstract idea. The claims are not patent eligible. With regards to claims 10 / 19, the claims recite the further limitation of "at least one of the computational paths entering one of the at least two hidden nodes can further activate one or more of the plurality of nodes positioned closer to the at least one output node". This further describes continuing the path traversal and marking downstream nodes as active. That is a further step of evaluating a graph and deciding which nodes to activate, which could be performed by a human using paper, pen and calculator, and a further step of the mathematical sampling. It therefore falls within the "Mental Processes" and "Mathematical Concepts" groupings of abstract ideas. Except for citing generic computer elements to implement the abstract idea, there is no additional element showing integration into a practical application or adding something significantly more to the abstract idea. The claims are not patent eligible. With regards to claim 11, the claim recites the further limitation of "each of the plurality of nodes acts independently to balance its probability of activation with the connected nodes". This describes a mathematical adjustment of probability values among connected nodes (¶ [0041]). It therefore falls within the "Mathematical Concepts" grouping of abstract ideas. Except for citing generic computer elements to implement the abstract idea, there is no additional element showing integration into a practical application or adding something significantly more to the abstract idea. The claim is not patent eligible. With regards to claims 12 / 20, the claims recite the further limitations of "the plurality of computational paths comprises at least one previously activated path, wherein the at least one previously activated path comprises a value …." (mental process and/or math relation), which is further process of evaluating data in a graph model, which could also be performed by a human using paper, pen and calculator, by writing down an intermediate result and referring to it later. The claim recites the additional element of " wherein the value of the at least one previously activated path is stored for use by one or more of the plurality of computational paths", which is extra-solution activity of pre-solution data input and/or post-solution data output (MPEP 2106.05(g)). Accordingly, at Step 2A, prong two, the additional elements individually or in combination do not integrate the judicial exception into a practical application. The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above, the additional element of “wherein the value of the at least one previously activated path is stored for use by one or more of the plurality of computational paths”, which is which is extra-solution activity of pre-solution data input and/or post-solution data output (see MPEP 2016.05(g)). The courts have found limitations directed to obtaining or transmitting information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory"). Accordingly, at Step 2B the additional elements individually or in combination do not amount to significantly more than the judicial exception. Accordingly, at Step 2B the additional elements individually or in combination do not amount to significantly more than the judicial exception. The claim is not patent eligible. With regards to claim 21, the claim recites the further limitations of "the selective activation of a subset of the plurality of nodes is scheduled according to a sampled or expected delay for at least one of the plurality of computational paths", and "processing performed on a first-in first-out basis regardless of a computed wait time". Drawing a sampled or expected delay is a mathematical calculation (¶¶ [0037], [0043]), and processing items in first-in, first-out order is an ordering of events that could be performed by a human using paper, pen and calculator. These limitations therefore fall within the "Mathematical Concepts" and "Mental Processes" groupings of abstract ideas. The claim recites the additional element of "the sampled or expected delay is stored in a buffer", which is extra-solution activity of pre-solution data input and/or post-solution data output (MPEP 2106.05(g)). Accordingly, at Step 2A, prong two, the additional elements individually or in combination do not integrate the judicial exception into a practical application. The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above, the additional element of “the sampled or expected delay is stored in a buffer”, which is which is extra-solution activity of pre-solution data input and/or post-solution data output (see MPEP 2016.05(g)). The courts have found limitations directed to obtaining or transmitting information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory"). Accordingly, at Step 2B the additional elements individually or in combination do not amount to significantly more than the judicial exception. Accordingly, at Step 2B the additional elements individually or in combination do not amount to significantly more than the judicial exception. The claim is not patent eligible. 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 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 1-5, 8-10, 12-15 and 17-20 are rejected under 35 U.S.C. 103 as being unpatentable over David et al., US-PGPUB NO. 20200279167A1 [hereafter David] in view of Bengio et al., "Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computation," arXiv:1308.3432, 2013 [hereafter Bengio]. With regards to claim 1, David teaches: "A computing system, comprising: (David, [0045], [0053], FIG. 3, 'a system 300 for partially-activating a neural network using a plurality of M partial pathways'; '… one or more controller(s) or processor(s) 314 and 324 … for executing operations … and one or more memory unit(s) 316 and 326', PNG media_image1.png 308 524 media_image1.png Greyscale )" " a … neural network comprising: a plurality of nodes; and (David, [0023], [0031], FIG. 1-2, 'Neural network 100 includes a plurality of artificial neurons 102', PNG media_image2.png 357 759 media_image2.png Greyscale )" "a plurality of synapses, wherein each synapse of the plurality of synapses connects a respective pair of the plurality of nodes; (David, [0031], FIG. 1-2, 'Each neuron 102 in each layer is connected by an artificial synapse connection 104 to some neurons (partially-connected) or to all neurons (fully-connected) in an adjacent layer', each synapse 104 connects a pair of neurons 102)" "wherein a subset of the plurality of nodes is selectively activated based on sampling a plurality of computational paths traversing the … neural network. (David, FIG.1-2, 4, [0017], [0022], [0036], [0057]–[0058], FIGS. 1B, 1C and 4, operations 410–420): [0017]: 'activating only a subset of neurons or filters along select pathways across a portion of the neural network multiple times. The multiple partial pathways may be selected either fully at random, or semi-randomly'; [0022]: 'A Monte Carlo or other random sampling technique may be used'; [0036]: 'different respective sequences of synapses continuously linking a single neuron in an input layer … to a single neuron in an output layer'; [0058]: 'partially-activate the neural network by iteratively executing a plurality of M partial pathways of the neural network to generate M partial outputs'. PNG media_image3.png 483 589 media_image3.png Greyscale Each partial pathway is a computational path traversing the network from input layer to output layer. The pathways are selected by random or semi-random (Monte Carlo) sampling. Only the neurons along the sampled pathways are activated, so a subset of the nodes is selectively activated based on sampling a plurality of computational paths.)" David does not explicitly detail a "stochastic neural network" in which the nodes themselves are activated selectively stochastically, i.e., fire with a probability that depends on their inputs.” However, Bengio teaches a "stochastic neural network" (Bengio, FIG.1, PNG media_image4.png 316 668 media_image4.png Greyscale Section 1 Introduction and Background, stochastic binary neurons that 'output a 1 or a 0 according to a sigmoid probability'; Section 3 Unbiased Estimator of Gradient for Stochastic Binary Neurons, Eq. 3, each neuron's binary output is drawn according to the sigmoid of its weighted input; Section 5 Conditional Computation Experiments, such stochastic neurons are used as gating units that select which hidden units of the network are actually computed for a given input). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the teachings of David and Bengio before him or her, to modify the partially-activated neural network of David to include a stochastic neural network as shown in Bengio. The motivation for doing so would have been for conditional computation (Bengio, Abstract). With regards to claim 13, David teaches: "A method for partially or selectively activating a … neural network, comprising: (David, [0009], [0055], FIG. 4, 'there is provided a device, system and method for approximating a neural network … The neural network may be partially-activated')" "providing a … neural network comprising a plurality of nodes; (David, [0056], [0031], FIG. 4, operation 400, 'a processor may store a neural network (e.g., in FIG. 1A or 200 in FIG. 2) comprising a total of N synapses or filters in a memory'; 'Neural network 100 includes a plurality of artificial neurons 102', storing the network in memory is providing it)" "providing a plurality of synapses, wherein each of the plurality of synapses comprises a connection between a respective pair of the plurality of nodes; (David, [0031], FIG. 1A, 'Each neuron 102 in each layer is connected by an artificial synapse connection 104 to some neurons … in an adjacent layer')" "providing a plurality of activation weights, wherein each of the plurality of activation weights is associated with a respective synapse of the plurality of synapses or a respective node of the plurality of nodes; and (David, [0031], [0038], 'Neural network 100 may be represented by a plurality of weights representing the strengths of the respective plurality of synapse connections 104'; 'Selection factors may include, for example, the strength of the various synapse weights (e.g., favoring synapses with stronger weights)', each synapse has a weight that governs whether the pathway, and hence the downstream neuron, is activated)" "selectively activating a subset of the plurality of nodes based on sampling a plurality of computational paths traversing the … neural network. (David, [0017], [0022], [0036], [0057]–[0058], FIGS. 1B, 1C and 4, operations 410–420, as quoted and explained for claim 1)" David does not explicitly detail a "stochastic neural network" whose nodes are selectively activated stochastically according to their weighted inputs.” However, Bengio teaches a "stochastic neural network" (Bengio, Section Introduction and Background; Section 2, Non-Smooth Stochastic Neurons, each neuron's input is an affine transformation of its incoming weights and bias; Section 3 Straight-Through Estimator, Eq. 3, the binary output is drawn according to the sigmoid of that weighted input, so the incoming weights act as activation weights). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the teachings of David and Bengio before him or her, to modify the method of David to to include a stochastic neural network as shown in Bengio. The motivation for doing so would have been for conditional computation (Bengio, Abstract). With regards to claim 2, David in view of Bengio teaches the limitations of claim 1, and further teaches: "wherein each of the plurality of computational paths selectively activates a sequence of nodes, (David, [0026], [0036], 'a "linear" pathway comprises a chain of neurons connected one-to-one'; 'sequences of synapses continuously linking a single neuron in an input layer … to a single neuron in an output layer', each pathway activates a sequence of neurons)." David does not explicitly detail a " wherein a computed output of each node in the sequence of nodes is dependent upon the probability of such a node being activated for a set of input values.” However, Bengio teaches a "wherein a computed output of each node in the sequence of nodes is dependent upon the probability of such a node being activated for a set of input values. (Bengio, Section 2 Non-Smooth Stochastic Neurons, Eq. 2, p. 5, the output of the stochastic unit is computed from its activation probability, i.e., the sigmoid of the unit's input, so the output depends on the probability of the unit being activated for the given input).”. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the teachings of David and Bengio before him or her, to modify the partially-activated neural network of David to include a stochastic neural network as shown in Bengio. The motivation for doing so would have been for conditional computation (Bengio, Abstract). With regards to claims 3 and 14, David in view of Bengio teaches the limitations of claims 1 and 13, respectively, and David further teaches: "wherein an output of the stochastic neural network comprises a sequence of the sampled plurality of computational paths with a corresponding sequence of output values that represent approximations of the output of the stochastic neural network. (David, FIG.4, [0058], [0060], [0021], [0066]): [0058]: 'iteratively executing a plurality of M partial pathways of the neural network to generate M partial outputs'; [0060]: 'the first stage may generate a rough solution that is further and iteratively refined by each subsequent stage'; [0021]: 'more partial runs of a neural network may be performed in a fixed time, and thus the accuracy of the output may be improved'; [0066]: 'aggregate the M partial outputs to generate an aggregated partial-activation output approximating a full-activation output'. The iteratively executed pathways produce a sequence of partial outputs, each an approximation of the full network output that is progressively refined.)" With regards to claim 4, David in view of Bengio teaches the limitations of claim 3, and David further teaches: "further comprising programming operable by the stochastic neural network to cease the sampling of the plurality of computational paths once the output of the stochastic neural network is sufficiently precise. (David, FIG.3-4, [0021], [0075]): [0021]: 'a variable number M of the partial pathways may be iteratively executed until the prediction results converge or the accuracy of the training results are above a predetermined threshold'; [0075]: 'a non-transitory computer or processor readable medium … storing instructions … which, when executed by a processor … carry out methods disclosed herein'. Iteratively executing pathways until the results converge is ceasing the sampling once the output is sufficiently precise, and the stored instructions are the programming.)" With regards to claims 5 and 15, David in view of Bengio teaches the limitations of claims 1 and 13, respectively, and David further teaches: "wherein the plurality of nodes comprises at least one input node, at least one output node and at least two hidden nodes, wherein the at least two hidden nodes are positioned between the at least one input node and the at least one output node in the stochastic neural network, (David, FIG. 1-2, PNG media_image5.png 436 261 media_image5.png Greyscale [0031], 'Neural network 100 may include an input layer (neurons 1-4), one or more middle or hidden layer(s) (neurons 5-8), and an output layer (neurons 9-12)')" "and wherein sampling the plurality of computational paths comprises initiating each of the plurality of computational paths from a first of the at least two hidden nodes, wherein the first of the at least two hidden nodes has been activated by a previous computational path. (David, [0074], [0037]): [0074]: 'the first set of layers execute using all of the neurons in those layers, then the result from the last such layer can be saved and used as input to all of the partial pathway runs of the rest of the network'; [0037]: 'partial pathways may connect specific layers (e.g., deeper middle layers …)'. The partial pathway runs are initiated from hidden-layer neurons whose values were already computed and activated by a prior computation through the earlier layers, i.e., by a previous computational path, rather than being re-sampled from the input nodes.)" With regards to claims 8 and 17, David in view of Bengio teaches the limitations of claims 1 and 13, respectively, and David further teaches: "wherein the plurality of computational paths is sampled in parallel (David, FIG.1-2, 4, [0027], [0059], 'the M partial pathways of the neural network may be executed in sequence (one pathway executed at a time), in parallel (all M partial pathways executed simultaneously), or in a semi-parallel manner')." With regards to claims 9 and 18, David in view of Bengio teaches the limitations of claims 1 and 13, respectively, and further teaches: "wherein the plurality of computational paths is sampled independently by each node or each synapse. (David, FIG.1-2, 4, [0038], 'Subsequent neurons or synapses in the first partial pathway may be selected fully-randomly or semi-randomly', the random selection is made at each neuron or synapse along the pathway)." With regards to claims 10 and 19, David in view of Bengio teaches the limitations of claims 5 and 15, respectively, and David further teaches: "wherein at least one of the computational paths entering one of the at least two hidden nodes can further activate one or more of the plurality of nodes positioned closer to the at least one output node. (David, FIGS.1-2, 4, [0036], [0074] 'sequences of synapses continuously linking a single neuron in an input layer … to a single neuron in an output layer'; 'used as input to all of the partial pathway runs of the rest of the network', a pathway entering a hidden neuron continues through subsequent layers to the output layer, activating the neurons closer to the output)." With regards to claims 12 and 20, David in view of Bengio teaches the limitations of claims 1 and 15, respectively, and David further teaches: "wherein the plurality of computational paths comprises at least one previously activated path, wherein the at least one previously activated path comprises a value, and wherein the value of the at least one previously activated path is stored for use by one or more of the plurality of computational paths. (David, FIGS.1-2, 4, [0060], [0062], [0074]): [0060]: 'The partial outputs of executing partial pathways in a relatively earlier stage … may be used to determine which partial pathways to execute in a subsequent stage'; [0062]: 'the processor may measure the accuracy of each pathway's output to determine or rank the performance of the pathways. The processor may implement a feedback loop, such that, pathways with higher performance ranking may be selected preferentially'; [0074]: 'the result from the last such layer can be saved and used as input to all of the partial pathway runs'. The outputs of previously executed pathways are retained and used by subsequently executed pathways.)" Additional Relevant Art The prior art made of record is considered pertinent to applicant’s disclosure and is recorded on Form PTO-892. Applicant is required under 37 C.F.R. § 1.111 (c) to consider these references fully when responding to this action, with particular attention paid to: Sridharan et al., US-PGPUB NO.20180322386A1 [hereafter Sridharan] shows partial activation of neural networks. Sridharan1 et al., US-PGPUB NO.20190205745A1 [hereafter Sridharan1] shows distributed training of neural networks. Conclusion Any inquiry concerning this communication or earlier communications from the examiner should be directed to TSU-CHANG LEE whose telephone number is 571-272-3567. The fax number is 571-273-3567. If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Omar Fernandez Rivas, can be reached 571-272-2589. Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000. /TSU-CHANG LEE/ Primary Examiner, Art Unit 2128
Read full office action

Prosecution Timeline

May 03, 2024
Application Filed
Oct 02, 2026
Non-Final Rejection mailed — §101, §103, §112 (current)

Precedent Cases

Applications granted by this same examiner with similar technology

Patent 12755177
COMPUTATION DEVICE, QUANTUM STATE GENERATION DEVICE, COMPUTATION METHOD, QUANTUM STATE GENERATION METHOD, AND PROGRAM
3y 2m to grant Granted Oct 06, 2026
Patent 12755213
HYBRID LANGUAGE MODEL AND DETERMINISTIC PROCESSING FOR UNCERTAINTY ANALYSIS
1y 2m to grant Granted Oct 06, 2026
Patent 12743608
NEURON CIRCUITS FOR A SPIKING NEURAL NETWORK BASED ON MAGNETIC-TUNNEL-JUNCTION LAYER STACKS CONNECTED IN SERIES
2y 11m to grant Granted Sep 22, 2026
Patent 12737623
REVERSE DATA GENERATION AND DATA DISTRIBUTION ANALYSIS TO VALIDATE ARTIFICIAL INTELLIGENCE MODEL
3y 5m to grant Granted Sep 15, 2026
Patent 12731017
CONIC PROGRAMMING BASED ON NEURAL NETWORK WITH SPIKING NEUROMORPHIC ARCHITECTURE
3y 4m to grant Granted Sep 08, 2026
Study what changed to get past this examiner. Based on 5 most recent grants.

Strategy Recommendation AI-generated — please review before filing

Get a prosecution strategy drawn from examiner precedents, rejection analysis, and claim mapping.
Typically takes 5-10 seconds — AI-generated, attorney review required before filing

Prosecution Projections

1-2
Expected OA Rounds
73%
Grant Probability
86%
With Interview (+13.6%)
3y 6m (~1y 0m remaining)
Median Time to Grant
Low
PTA Risk
Based on 434 resolved cases by this examiner. Grant probability derived from career allowance rate.

Sign in with your work email

Enter your email to receive a magic link. No password needed.

Personal email addresses (Gmail, Yahoo, etc.) are not accepted.

Free tier: 3 strategy analyses per month