DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA . This action is in response to a restriction made to the case on January 27th, 2026. Claims 8-20 are pending in the current application.
Claim Objections
Claims 13, 14, and 18 is objected to because of the following informalities: The word “randomised” of claim 13 and variations of the word “vectorise” (“vectorized” and “vectorizing”) of claims 13, 14, and 18 are misspelled. They should be “randomized” and “vectorize” respectively. 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.
Claims 11, 12, 14, 16, 18, and 20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
The claims are replete with issues that render the claims indefinite. Some examples include, but are not limited to:
Claim 14 recites “performing a minor mutation” and “performing a major mutation. The terms “minor” and “major” is a relative term which renders the claim indefinite. The terms “minor” and “major” is not defined by the claim, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention. The term renders the scope of the mutation indefinite.
Claims 11, 12, 16, 18, and 20 recite optional steps. It is unclear whether the steps are required to be performed as a part of the inventive concept, rendering the scope of the invention indefinite.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 8-12, 18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Abdelhamid Zemirline et al. (Herein referred to as Zemirline) (Rule-based diagnostic system fusion) in view of Zhao Zhang et al. (Herein referred to as Zhang) (Relational Graph Neural Network with
Hierarchical Attention for Knowledge Graph Completion) and in further view of Sumit Pai et al. (Herein referred to as Pai) (Learning Embeddings from Knowledge Graphs With Numeric Edge Attributes)
Regarding claim 8, Zemirline teaches a computer implemented method of fusing or integrating at least two agent model(s) for modelling a complex system, each agent model comprising: a plurality of agent system, AS, node(s), wherein each of the AS node(s) comprise a plurality of agent units, AUs, (“our fusion method is a model fusion because it combines knowledge bases and rule bases of several diagnostic systems in order to build a new diagnostic system.”, pg. 1, right column, second paragraph after the bulleted list)) (Knowledge bases inherently contain nodes, which contains units and rules that make up units.) a set of AS rules governing the plurality of AUs (“This rule set contains the rules of each class and the diagnostic system uses the best rules of each class for prediction [7].”, pg. 3, right column, first paragraph) wherein each agent model is trained to model one or more portion(s) of the complex system using a corresponding labelled training dataset (“In this section, we evaluated our knowledge fusion approach on four databases from an endoscopic image analysis system… In all the bases, there are 206 parameters and 89 types of lesions, (i.e. 89 label classes).”, pg. 7, right column, under “IV. Evaluation”) said each agent model being adapted, during training, to form: an agent rule base comprising one or more sets of AS rules (“We note that the diagnostic system based on the training basis A has the higher rate of accuracy than other systems based on other training bases”, pg. 7, right column, second to last paragraph; See also Table I on pg. 8) (For a training database, a number of rules are established, corresponding to an agent rule base.) the method comprising: determining an intersecting rule set between the agent rule bases of at least a first trained agent model and a second trained agent model (“This method is applied to the diagnostic system rule bases. It aims at combining all the rule bases into only one rule base and then taking into consideration the characteristics of this base. The fusion method is characterized by a hybrid fusion which combines rule fusion approach with knowledge fusion approach.”, pg. 2, Abstract) (The combined rule bases correspond to a intersecting rule set) merging said at least first and second trained agent models to form an integrated agent model based on combining those one or more layer(s), AS node(s), and/or AU(s) of the first and second trained agent models that correspond to the intersecting rule set (“Besides, our fusion method is a model fusion because it combines knowledge bases and rule bases of several diagnostic systems in order to build a new diagnostic system.”, pg. 2, right column, fourth paragraph) (Model fusion merges two models, which includes the layers, nodes, or units. As Zemirline also combines rule bases, the model would correspond to the intersecting rule base.)
However, Zemirline does not explicitly teach an input layer comprising a set of AS nodes of the plurality of AS node(s); an output layer comprising at least one AS node of the plurality of AS node(s); and one or more intermediate layer(s), each of the intermediate layer(s) comprising another set of AS node(s) of the plurality of AS node(s); nor an agent network state comprising data representative of the interconnections between the AS nodes of the input, output and intermediate layer(s), wherein the agent rule base and agent network state are generated during training and configured for modelling said portion(s) of the complex system; nor updating the integrated agent model based on one or more validation and training labelled datasets associated with each of the at least first and second trained agent model(s) until the integrated model is validly trained.
Zhang teaches an input layer comprising a set of AS nodes of the plurality of AS node(s) an output layer comprising at least one AS node of the plurality of AS node(s) (“For the encoder, the embedding size of entities is set as 100 for both the input and output layer. The number of heads for the multi-head attention mechanism is set as 8. And the bi-interaction combination is utilized by the information aggregator. In addition, a dropout with the rate as 0.5 is applied to each input layer of the encoder and the normalized attention coefficients following graph attention network”, pg. 5, right column, under “Experimental Settings”) (In combination with the nodes of Zemirline, the input and output layer of Zhang comprises a set of nodes.) And an agent network state comprising data representative of the interconnections between the AS nodes of the input, output and intermediate layer(s), wherein the agent rule base and agent network state are generated during training and configured for modelling said portion(s) of the complex system (“In the training stage, we adopt a two-layer RGHAT to train the entity and relation embeddings. For the encoder, the embedding size of entities is set as 100 for both the input and output layer.”, pg. 5, right column, under “Experimental Settings”) (The entity embeddings corresponds to data representative of the interconnections between the nodes of the different layers) and updating the integrated agent model based on one or more validation and training labelled datasets associated with each of the at least first and second trained agent model(s) until the integrated model is validly trained. (“…each entity aggregates information and gets updated from its neighborhood based on the hierarchical attentions… In the training stage, we adopt a two-layer RGHAT to train the entity and relation embeddings”, pg. 2, left column, first paragraph; pg. 5, right column, under “Experimental Settings”; See also Table 1 on pg. 5) (The model uses validation data which corresponds to each dataset according to Table 1, corresponding to validation data associated with each of the models. During the training process, the model is updated based on the attention from the dataset associated neighborhood.)
Therefore, it would have been considered obvious to one of ordinary skill in the art, prior to the filing date of the current application, to combine the model and knowledge fusion system of Zemirline with the input and output layers, as well as the updating step using a validation dataset of Zhang. One would have been motivated to combine the teachings, prior to the application’s filing date, as Zhang’s method has consistency with neighborhood aware models. (“Moreover, further analysis showed the results of RGHAT were more consistent
with human intuition compared to other neighborhood-aware models. Finally, extensive experiments on popular benchmarks clearly validated the superiority of RGHAT against various state-of-the-art baselines.”, pg. 8, left column, first paragraph; See also pgs. 6 and 7 for specific metrics)
However, the combination still does not explicitly teach one or more intermediate layer(s), each of the intermediate layer(s) comprising another set of AS node(s) of the plurality of AS node(s)
Pai teach one or more intermediate layer(s), each of the intermediate layer(s) comprising another set of AS node(s) of the plurality of AS node(s) (“The add-on acts as an intermediate layer between the traditional scoring layer and the loss.”, pg. 3, Figure 2)
Therefore, it would have been considered obvious to one of ordinary skill in the art, prior to the filing date of the current application, to combine the fusion system of Zemirline, as modified by Zhang, with the add-on intermediate layer of Pai. One would have been motivated to combine the two teachings, prior to the application’s filing date, as Pai’s intermediate layer shows better capabilities of ranking high-valued triples. (“FocusE brings larger differences in MRR across the board, showing better capabilities at correctly ranking the top-10% high-valued triples vs the bottom 10%. Best results in bold.”, pg. 6, bottom, See Table 4)
Regarding claim 9, Zemirline, as modified by Zhang and Pai, teaches the computer implemented method according to claim 8, wherein the complex system is modelled by a plurality of agent model(s), each agent model of the plurality of agent model(s) configured to model a different portion of the complex system, the method further comprising: determining an intersecting rule set between two or more of the agent rule bases of the plurality of agent model(s), wherein the agent rule base of each of the plurality of agent model(s) intersects with at least one other agent rule base of another of the plurality of agent model(s); (“This method is applied to the diagnostic system rule bases. It aims at combining all the rule bases into only one rule base and then taking into consideration the characteristics of this base. The fusion method is characterized by a hybrid fusion which combines rule fusion approach with knowledge fusion approach.”, pg. 2, Abstract (Zemirline)) (The combined rule bases correspond to an intersecting rule set. This can be repeated to work with any other rule agent bases.) And merging said plurality of agent models to form an integrated agent model based on combining those one or more layer(s), AS node(s), and/or AU(s) of each of the plurality of agent models that correspond to the intersecting rule set; (“Besides, our fusion method is a model fusion because it combines knowledge bases and rule bases of several diagnostic systems in order to build a new diagnostic system.”, pg. 2, right column, fourth paragraph (Zemirline)) (Model fusion merges two models, which includes the layers, nodes, or units. As Zemirline combines rule bases, the model would then correspond to the intersecting rule base.) And updating the integrated agent model based on one or more validation and training labelled datasets associated with each of the at least first and second trained agent model(s) until the integrated model is validly trained. (“…each entity aggregates information and gets updated from its neighborhood based on the hierarchical attentions… In the training stage, we adopt a two-layer RGHAT to train the entity and relation embeddings”, pg. 2, left column, first paragraph; pg. 5, right column, under “Experimental Settings”; See also Table 1 on pg. 5 (Zhang)) (The model uses validation data which corresponds to each dataset according to Table 1, corresponding to validation data associated with each of the models. During the training process, the model is updated based on the attention from the dataset associated neighborhood.)
Regarding claim 10, Zemirline, as modified by Zhang and Pai, teaches the computer implemented method according to claim 8 wherein the complex system is modelled by a plurality of agent model(s), each agent model of the plurality of agent model(s) configured to model a different portion of the complex system, the method further comprising: for each agent model of the plurality of agent models, determining an intersecting rule set between the agent rule bases of said each agent model and any of those other agent models in the plurality of agent model(s), wherein the agent rule base of each of the plurality of agent model(s) intersects with at least one other agent rule base of another of the plurality of agent model(s); (“This method is applied to the diagnostic system rule bases. It aims at combining all the rule bases into only one rule base and then taking into consideration the characteristics of this base. The fusion method is characterized by a hybrid fusion which combines rule fusion approach with knowledge fusion approach.”, pg. 2, Abstract (Zemirline)) (The combined rule bases correspond to an intersecting rule set. This can be repeated to work with any other rule agent bases.) And for each agent model of the plurality of agent models, merging said each agent model with each of those agent models in the plurality of agent models determined to intersect with said each agent model to form an intermediate fused or integrated agent model based on combining those one or more layer(s), AS node(s), and/or AU(s) of each of the plurality of agent models that intersect; (“Besides, our fusion method is a model fusion because it combines knowledge bases and rule bases of several diagnostic systems in order to build a new diagnostic system.”, pg. 2, right column, fourth paragraph (Zemirline)) (Model fusion merges two models, which includes the layers, nodes, or units.) merging each of the intermediate fused or integrated agent models to form an fusion agent model; (“Besides, our fusion method is a model fusion because it combines knowledge bases and rule bases of several diagnostic systems in order to build a new diagnostic system.”, pg. 2, right column, fourth paragraph (Zemirline)) (Model fusion merges two models, which includes the layers, nodes, or units. Zemirline’s fusion method can work with any knowledge model, including intermediate fused or integrated agent models.) and updating the fusion agent model based on one or more validation and training labelled datasets associated with each of the plurality of agent models until the integrated model is validly trained. (“…each entity aggregates information and gets updated from its neighborhood based on the hierarchical attentions… In the training stage, we adopt a two-layer RGHAT to train the entity and relation embeddings”, pg. 2, left column, first paragraph; pg. 5, right column, under “Experimental Settings”; See also Table 1 on pg. 5 (Zhang)) (The model uses validation data which corresponds to each dataset according to Table 1, corresponding to validation data associated with each of the models. During the training process, the model is updated based on the attention from the dataset associated neighborhood.)
Regarding claim 11, Zemirline, as modified by Zhang and Pai, teaches The computer implemented method according to claim 8, wherein determining an intersecting rule set between at least the first trained agent model and second trained agent model further includes determining a compatibility score between at least the first trained agent model and the second trained agent model, and indicating those models of at least the first and second trained agent models to be merged when the compatibility score is above a predetermined threshold; (“We define a confidence measurement function of all parameters in the new base: where ε ∈]0,1], which is used as threshold in order to estimate the confidence of a parameter from the conflict value.”, pg. 6, left column, second to bottom paragraph (Zemirline)) (The confidence measurement corresponds to a compatibility score.) optionally wherein calculating the compatibility score comprises determining whether one or more semantic relationships exist between at least the first trained model and at least the second trained model; (“In the training stage, we adopt a two-layer RGHAT to train the entity and relation embeddings.”, pg. 5, right column, under “Experimental Settings”; See also Table 7 and Table 8 on pg. 7. (Zhang)) (This limitation is recited as an optional limitation, and is such given negligible patentable weight. Even if it was written to positively recite this limitation, Zhang’s relation embedding corresponds to a determination of whether one or more semantic relationships exist. In combination with the multiple models and knowledge bases of Zemirline, the limitation is fully taught.) optionally wherein determining whether one or more sematic relationships exist further comprises forming a semantic network between at least the first trained model and the second trained model, wherein interconnections in the semantic network exist when one or more entities associated with the first trained model are connected, correlate or have a relationship with one or more entities associated with the second trained model. (“The entity-level attention is inspired by the intuition that the weights of neighboring entities under the same relation may also be different. For instance, the relation has players links Los Angeles Lakers to different players, among all these players, top stars may be more indicative than other players. The proposed model first views the neighboring entities under the same relation as a group, then computes the entity-level attention”, pg. 3, right column, under “Entity-level Attention” (Zhang)) (This limitation is recited as an optional limitation, and is such given negligible patentable weight. Even if it was written to positively recite this limitation, the relations of Zhang form a network with one another, as shown in Figure 3. Alongside the multiple model of Zemirline, the limitation is fully taught.)
Regarding claim 12, Zemirline, as modified by Zhang and Pai, teaches The computer implemented method according to claim 8 the steps of determining an intersection rule set and merging at least the first trained agent model and at least the second trained agent model further comprising: determining one or more areas of similarity between agent state networks of at least the first trained agent model and second trained agent model; (“…importance, e.g., in the results of Nathani’s, the weights of triple (Los Angeles Lakers, sports team roster/player, Magic Johnson) and (Los Angeles Lakers, sports team roster/player, Kobe Bryant) are 0.0759 and 0.0108, respectively. The former is almost 7 times of the latter, which is obviously inconsistent with our intuition. Our model assigns the weights of 0.0664 and 0.0617 to the two triples (see Table 7), which is considered to be more suitable and is consistent with our intuition.”, pg. 7, right column, second paragraph; See also Table 7 and 8 (Zemirline)) (The importance between the head and tail entities correspond to a determination in an area of similarity. Combined with the model and knowledge fusion of Zemirline, the limitation is fully taught.) comparing, based on each area of similarity, the AS rule sets of the AS nodes in the area of similarity between at least the first trained agent model and the second trained agent model; (“The external knowledge base is the representation of the case base characteristics. It is considered as an interface that permits to compare the rule bases of various systems… Compare the expected average accuracy of the best k rules of each class and choose the class with the highest expected accuracy as the prediction class.”, pg. 3, left column, under “II. Diagnosis System”; pg. 3, right column, number 3 on the numerical list (Zemirline)) and merging, based on the comparison of each area of similarity, the corresponding AS nodes and interconnections between the layers of at least the first and second trained models; (“Besides, our fusion method is a model fusion because it combines knowledge bases and rule bases of several diagnostic systems in order to build a new diagnostic system.”, pg. 2, right column, fourth paragraph (Zemirline)) (Model fusion merges two models, which includes the layers, nodes, or units. The rules are merged and compared to determine the best rules of each class.) optionally wherein the method further comprises: determining, using a graph matching algorithm, the one or more areas of similarity between at least the first trained agent model and the second trained agent model; and merging the corresponding AS nodes and interconnections further comprising: concatenating, based on the determined areas of similarity, the corresponding sets of AS rules and AS node states of the at least first trained agent model and the second trained agent model; and applying a belief function to the concatenated set of AS rules and AS states. (The limitation, as drafted, recites an optional step, and as such is given negligible patentable weight. If the step was written to be more positively recited, the current art would likely not cover the currently optional limitations)
Regarding claim 18, Zemirline, as modified by Zhang and Pai teaches The computer implemented method according to claim 8, wherein each agent of the plurality of agent(s) of an AS node has a local state vector; (“The information aggregator aggregates information from the local neighborhood to the central entity, and gets the neighborhood-based representation for entity h,”, pg. 4, left column, first paragraph (Zemirline)) (The central entity corresponds to a local state vector.) optionally wherein a value of the local state vector is updated after an iteration of a cycle and/or is set based on a historical value; optionally further comprising: processing, the vectorized data at the input layer comprises: determining a firing threshold at the input AS node based on the received one- dimensional data; computing, at the input AS node in the input layer, a transformation of the one- dimensional input vector to a first vector of first size based on the firing threshold of the input AS node; and transmitting or propagating, from the input layer to the one or more intermediate layer, the first vector to each agent of the plurality of AS nodes. (These limitations are recited to be optional, and therefore given negligible patentable weight. If more positively recited, the limitation may then require further search.)
Regarding claim 20, Zemirline, as modified by Zhang and Pai, teaches the computer implemented method according to claim 8, wherein each AS node of the one or more intermediate and output layer(s) is coupled to a select/reduce function component configured for receiving each of the one or more output vectors from one or more upstream AS node(s) connected to said each AS node, wherein the select/reduce function component combines or transforms the received one or more output vectors into an input vector for input to said each AS node; (“The information aggregator aggregates information from the local neighborhood to the central entity and gets the neighborhood-based representation for entity h… Though ˆh incorporates the information from its local neighborhood, it lacks valuable information from itself. To this end, we further combine the neighborhood-based representation ˆh with the input representation h, and get the output representation h_.”, pg. 4, left column, first and second paragraph (Zhang)) (The aggregator, corresponding to a select/reduce function, aggregates information from the neighborhood of entities combined into a representation h^ which is then combined with an input representation to get a output representation. This eventually acts as input to a decoder, making the output correspond to an input vector to each node, fully teaching the limitation.)
Claim(s) 13, 15, 16 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Abdelhamid Zemirline et al. (Herein referred to as Zemirline) (Rule-based diagnostic system fusion) in view of Zhao Zhang et al. (Herein referred to as Zhang) (Relational Graph Neural Network with Hierarchical Attention for Knowledge Graph Completion) and in further view of Sumit Pai et al. (Herein referred to as Pai) (Learning Embeddings from Knowledge Graphs With Numeric Edge Attributes) and in further view of Mojtaba Nayyeri et al. Herein referred to as Nayyeri) (LogicENN: A Neural Based Knowledge Graphs Embedding Model with Logical Rules)
Regarding claim 13, Zemirline, as modified by Zhang and Pai, teaches The computer implemented method according to claim 8 further comprising: training each agent model further comprising: receiving each labelled training data item from a source and vectorizing each received labelled training data item (“The first set Etrain1 is made up of entities with top 10% degree values, the second set Etrain2 consists of entities with top 10% to 50% degree values, and the third set Etrain3 includes the remaining entities… In the training stage, we adopt a two-layer RGHAT to train the entity and relation embeddings. For the encoder, the embedding size of entities is set as 100 for both the input and output layer.”, pg. 5, left column, under Table 1; pg. 5, right column, under “Experimental Settings” (Zhang)) (The model is train off of the embeddings, which innately are vectors.) processing, by at least one of the input, intermediate and output layer(s), each vectorized training data item by the corresponding AS node(s), wherein the AS node(s) are located in the same or different layers and perform at least one of a plurality of functions; (“For entity h, the relation-level attention indicates the weight of each relation when representing the entity… where _ represents the concatenation operation, h ∈ Rd is the embedding of the entity h, and d is the embedding size. W1 ∈ Rd×2d, vr ∈ Rd and p ∈ Rd are training parameters, in which vr is a relation-specific parameter representing the characteristics of a relation. Nh represents the neighboring relations of entity h. σ represents the LeakyReLU function with negative input slope as 0.2. After the above calculation, we get the relation-level attention score αh,r, which represents the weight of relation r when representing entity h.”, pg. 3, right column, second paragraph, right above Equation 1 and below Equation 2) (The embedding are input to the model, and an attention score is generated.) outputting, from the output layer, an output vector for each labelled training data item in the labelled training dataset based on the processed vectorized training data item; (“A single-layer encoder aggregates information from 1-hop neighbors to the central entity in 1 training iteration. With the numbers of layers and iterations increase, our model can effectively aggregate information from multi-hop neighbors, which provide valuable information for representing the central entity. Finally, the encoder outputs the new entity embedding h_ to the decoder.”, pg. 4, right column, second paragraph (Zhang)) and updating the AS node(s) of at least one of the input, intermediate layer(s) based on comparing each output vector with each corresponding labelled training data item. (“…each entity aggregates information and gets updated from its neighborhood based on the hierarchical attentions… In the training stage, we adopt a two-layer RGHAT to train the entity and relation embeddings”, pg. 2, left column, first paragraph; pg. 5, right column, under “Experimental Settings”; See also Table 1 on pg. 5 (Zhang)) (When the model is updated, the layers are also updated.)
However the combination does not teach training each agent model to model one or more portions of the complex system using a labelled training dataset comprising a plurality of labelled training data items corresponding to the one or more portions of the complex system, wherein interconnections between AS nodes are initially randomized.
Nayyeri teaches training each agent model to model one or more portions of the complex system using a labelled training dataset comprising a plurality of labelled training data items corresponding to the one or more portions of the complex system, wherein interconnections between AS nodes are initially randomized (“The assumption is that there are global features which explain the existence of triples in a KG and embedding models try to capture those features using (typically low dimensional) vectors known as embeddings. Therefore, a KGE model assigns vectors (h,r,t) to the symbolic entities and relations (h, r,t). The vectors are initialized randomly and updated by solving an optimization problem.”, pg. 1, left column, bottom paragraph)
Therefore, it would have been considered obvious to one of ordinary skill in the art, prior to the filing date of the current application, to combine the model and knowledge fusion system of Zemirline, as modified by Zhang and Pai, with randomization and initialization of Nayyeri. One would have been motivated to combine the teachings, prior to the application’s filing date, as this allows for random initialization and the eventual updating of the embeddings to solve an optimization problem. (“Therefore, a KGE model assigns vectors (h,r,t) to the symbolic entities and relations (h, r,t). The vectors are initialized randomly and updated by solving an optimization problem., pg. 1, left and right columns)
Regarding claim 15, Zemirline, as modified by Zhang and Pai, teaches The computer implemented method according to claim 13 wherein, prior to processing one or more input vector(s), each AS node waits for all AS nodes connected to said each AS node to send the corresponding one of the one or more input vector(s), and process(es) said one or more input vector(s) once they all have been received. (“Information Aggregator The information aggregator aggregates information from the local neighborhood to the central entity, and gets the neighborhood-based representation for entity h… To this end, we further combine the neighborhood-based representation h^ with the input representation h, and get the output
representation h’.… In the final layer of the encoder, we average the embeddings from multiple heads instead of concatenating them…”, pg. 4, left column, first paragraph; pg. 4, right column, first paragraph (Zhang)) (The neighborhood-based representation, corresponding to a input vector is compiles via several nodes and their relationship to the central entity. Once the information is received, it can be combined with the input representation which waits for the neighborhood-based representation to be computed, teaching the limitation.)
Regarding claim 16, Zemirline, as modified by Zhang and Pai teaches The computer implemented method according to claim 13, further comprising, once an AS node sends the corresponding output vector towards one or more connected AS node(s), sending by said AS node the output vector to each one or more upstream AS node(s) connected to said AS node; (“Specifically, K dependent attention mechanisms calculate the embeddings, which are then concatenated, resulting in the following representation. where h_ k is the output representation of the k-th head.”, pg. 4, right column, second paragraph (Zhang)) (The last output corresponds to the output vector.) optionally wherein said each upstream AS node reduces a threshold for outputting an output vector. (This limitation is an optionally limitation, and as such given negligible patentable weight. If more positively recited, the limitation may then require further search.)
Regarding claim 19, Zemirline, as modified by Zhang and Pai teaches the computer implemented method according to claim 13, wherein vectorising the received data further comprises splicing the received data into a one-dimensional input vector based on one or more of: propagating the one-dimensional input vector to each AS node of the input layer; dividing the one-dimensional input vector into one or more portions, wherein each portion is propagated to a different AS node of the input layer; or applying a sliding window of a fixed length over the one-dimensional vector for propagating corresponding fixed length portions of the one-dimensional vector to a different AS node of the input layer. (“For the encoder, the embedding size of entities is set as 100 for both the input and output layer. The number of heads for the multi-head attention mechanism is set as 8. And the bi-interaction combination is utilized by the information aggregator. In addition, a dropout with the rate as 0.5 is applied to each input layer of the encoder and the normalized attention coefficients following graph attention network", pg. 5, under "Experimental Settings” (Zhang)) (Each of the entities receive the 100 dimensions of the input vector. It works similarly to Nayyeri’s Figure 1 on pg. 3)
Allowable Subject Matter
Claims 14 and 17 are seemingly allowable under prior art if written as independent claims. The claims may be fully allowable should the other rejections and objections made in this action are properly traversed.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Tyler E Iles whose telephone number is (571)272-5442. The examiner can normally be reached 9:00am - 5:00pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Kakali Chaki can be reached at (571) 272-3719. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/T.E.I./ Patent Examiner, Art Unit 2122
/MICHAEL H HOANG/PRIMARY EXAMINER, Art Unit 2122
Prosecution Insights