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 .
Status
This instant application No. 18/528,993 has claims 1-20 pending.
Priority / Filing Date
Applicant’s claims for priorities of parent applications 19/071,750 (now abandoned), 18/528,993 (now abandoned), No. 17/085,050 (now abandoned) and provisional application No. 62/928,893 are acknowledged. The effective filing date for this application is October 31, 2019.
Abstract
The abstract of the disclosure is objected due to the use of implied language. Note that in the abstract, the language should be clear and concise and should not repeat information given in the title. It should avoid using phrases which can be implied, such as, “The disclosure concerns,” “The disclosure defined by this invention,” “The disclosure describes,” etc… See MPEP § 608.01(b).
Note that in the abstract, Applicant cites “A method for performing hierarchical classification of data is disclosed” on line 1. This citation clearly provokes the use of implied language and repeats the title. Revision and/or correction are required (e.g., removal of the entire first sentence of the abstract).
Drawings
The drawings filed on October 6, 2025 are acceptable for examination purposes.
Information Disclosure Statement
As required by M.P.E.P. 609(C), the Applicant’s submission of the Information Disclosure Statement filed on October 6, 2025 is acknowledged by the Examiner and the cited references have been considered in the examination of the claims now pending. As required by M.P.E.P. 609 C(2), a copy of the PTOL-1449 initialed and dated by the Examiner is attached to the instant Office action.
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.
The claimed invention in claims 1-20 are directed to a judicial exception (i.e., an abstract idea) without significantly more.
a. Claims 1-20 pass step 1 of the 35 U.S.C. 101 analysis since each claim is either directed to a method, a system comprising a memory and at least one processing device coupled to the memory (i.e., hardware components [0029]-[0030] of specification), or a non-transitory computer-readable medium.
b. Claims 1, 9, and 18 recite in each, in part, elements that are directed to an abstract idea (“Courts have examined claims that required the use of a computer and still found that the underlying, patent-ineligible invention could be performed via pen and paper or in a person’s mind.” Versata Dev. Group v. SAP Am., Inc., 793 F.3d 1306, 1335, 115 USPQ2d 1681, 1702 (Fed. Cir. 2015)). Each claim recites the abstract idea of analyzing input information to identify attributes, temporal information, and relationships, and assigning the input information to one or mor hierarchical output classes. Such information analysis and classification constitutes an evaluation or judgment and is within the mental process grouping of abstract ideas per Step 2A – Prong 1. The claims additionally recite computer-implemented neural network processing for performing the information analysis and classification.
c. Per Step 2A – Prong 2
(i) Regarding claims 1, 9, and 18, the additional elements recited in each independent claim do not integrate the abstract idea into a practical application. Each independent claim recites, in substance:
one or more processing devices;
a memory or non-transitory computer-readable medium;
receiving input data;
encoding the data into multiple channels using deep CNN
extracting features;
extracting temporal structures;
identifying feature dependencies;
combining the extracted information into a combined feature set; and
classifying the combined feature set with a classifier having a plurality of nodes, each node comprising a NN.
These elements do not constitute a practical application of the identified abstract idea. The independent claims do not recite an improvement to the functioning of a computer, processor, memory, network ,database, or other computing technology. The claims also do not recite an improvement to another technology or technical field. Instead, the claims use a processing device and NN components as tools for receiving data, analyzing the data, and producing a classification result. Each of claims 1, 9, and 18 recites the desired results of encoding, extracting, identifying, combining, and classifying data without reciting how the NN components are technically configured to produce the asserted improvement of improving accuracy, flexibility, and robustness in hierarchical classification ([0051]-[0053] of specification). The recited deep CNN and NN classifier are therefore used as a generalized computational tools for carrying out the identified information analysis and classification concept. The claims do not effect a transformation or reduction of a particular article to a different state or thing. The recited manipulation of input data, feature information, temporal information, dependency information, and output classes is manipulation of information itself. The output is a classification of the received data, not a transformation of a physical article or technical action performed in a technological environment. As such, the additional elements of each independent claim, individually and in combination, do not integrate the abstract idea into a practical application.
(ii) Regarding claims 2-4, 10-12, and 19-20, they add a multilevel classifier, a hierarchical multilevel network of interconnected deep NN models, and/or training with training data associated with output classes.
These limitations do not integrate the abstract idea into a practical application because they continue to recite data analysis and hierarchical classification at a high level of generality. The claims do not specify the operative interconnections between the deep NN models, a model orchestration protocol, or a technical structure by which the hierarchical arrangement improves computing technology.
Reciting a multilevel or hierarchical classifier changes the organization of the classifications, but does not itself identify a particular technological application of the abstract information classification concept. Likewise, training a classifier with training data associated with output classes is conventional use of training data to configure a model for classification, and the claims do not recite a specific training technique or technical training data management mechanism that improves computer functionality.
(iii) Regarding claims 5-6, and 13-14, they add that the input data may comprise one or mor modalities corresponding to an input type and that the classifier facilitates taxonomy classification associated with the modalities.
These limitations do not integrate the abstract idea into a practical application since they broadly identify the type of information to be analyzed and the type of classification result to be generated. The claims do not require distinct modality specific classifiers, a shared class label structure, nor recite how data from different modalities is coordinated or processed. Therefore, the claims do not reflect a particular technical implementation of multimodal classification.
(iv) Regarding claims 7-8, and 15-16, they add that the input data comprises candidate profiles, resumes, songs, or movies, and/or that features include job skills or educational qualifications.
These limitations do not integrate the abstract idea into a practical application since they merely limit the content of the information being analyzed. The claimed result remains the analysis and classification of information. In particular, identifying job skills and educational qualifications in candidate profiles or resumes and assigning the profile or resume to a classification constitutes a field-of-use limitation directed to a particular type of data and does not improve a computer or other technology.
(v) Regarding claim 17, it adds classifying the data until a terminal class node is reached.
This limitation does not integrate the abstract idea into a practical application since it describes a result of progressing through a hierarchy of classifications until a final category is selected. The claim broadly recites hierarchical data classification and does not provide a technological application beyond the abstract classification concept.
d. Per Step 2B
The recited processing device, memory, input device, interface, display, and non-transitory medium are well-understood, routine, and conventional (WURC) computer components used according to their ordinary functions. The recited CNN, classifier, NN nodes, multilevel classifier, and training data likewise are invoked at a functional level to perform their WURC functions of encoding data, extracting information, determining relationships, combining information, and generating classifications.
When considered as an ordered combination, the claims recite a generalized ML classification workflow of receive input data → encode or represent the input data → extract features, temporal information, and dependencies → combine the extracted information → classify the information within one or more output classes. The claimed ordered combination does not add significantly more than the abstract idea because it merely automates information analysis and categorization using WURC computer and NN components. The claims do not recite a particular nonconventional arrangement by which feature-extraction, temporal-structure extraction, dependency identification, fusion, and hierarchical classification are performed. As such, the ordered combination therefore amounts to no more than implementation of the abstract information analysis and hierarchical classification concept on generic computing components using generalized NN tools.
Accordingly, claims 1-20 do not recite an inventive concept sufficient to transform the judicial exception into patent eligible subject matter.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-2, 4-6, 9-10, 12-14, and 18-19 are rejected under 35 U.S.C. 103 as being unpatentable over Amer et al. (Pub. No. US 2019/0304157, published on October 3, 2019; hereinafter Amer) in view of Singaraju et al. (Pub. No. US 2019/0103095, published on April 4, 2019; hereinafter Singaraju).
Regarding claims 1, 9, and 18, Amer clearly shows and discloses a method for performing hierarchical classification of data (Abstract); a system for performing hierarchical classification of data, the system comprising: a memory; and at least one processing device coupled to the memory, the at least one processing device is configured to implement the method; and a non-transitory computer-readable medium for storing instructions, wherein the instructions are executed by at least one processing device, wherein the at least one processing device is configured to implement the method (Figure 1C); wherein the method comprising:
receiving, by at least one processing device, input data (A CNN discriminator may still be used to encode the content of the input animations to provide context, [0056]);
encoding the input data into multiple channels comprising a convolutional deep neural network (the CNN Discriminator may mirror the CNN generator. The input video is first encoded into a N frames by k channels representation through size-1 one-dimensional convolution, [0058]);
extracting, by the at least one processing device, one or more features corresponding to the encoded input data (to detect an object, a CNN starts by detecting edges, combining them into more complex structures, such as parts in images, and then reasoning is done at the high-level representation, [0055]);
extracting, by the at least one processing device, one or more temporal structures of the encoded input data (each set of scenes or event sequences, information about a grounded data structure that would appropriately represent each set of scenes or event sequences included within training data 101. Data structure 151 corresponds to a representation of the scene or event sequences that includes each relevant spatio-temporal and object-attribute relationship, [0036]);
identifying, by the at least one processing device, one or more feature dependencies in the [encoded] input data, wherein the extracting steps and the identifying step are performed by the convolutional deep neural network (FIG. 2B illustrates parsed text 117′ being converted into a spatio-temporal composition graph 152′ (e.g., where composition graph 152′ an example composition graph 152 from FIG. 1C), with nodes for each of the actors (e.g., Neil and Cowboy), props (e.g., bathroom door, beer bottle), and locations (e.g., behind bar, next to chair) in the scene. Each actor is associated with a number of attributes (e.g., the Cowboy has a vest, a hat, and is tall), [0102]);
combining, by the at least one processing device, the extracted one or more features corresponding to the encoded input data, the extracted one or more temporal structures of the input data, and the identified one or more feature dependencies in the encoded input data, into a combined feature set (Such an event frame may be in the form of a “who” “did what” “to whom” “where” and “when” format. Accordingly, graph module 322 may identify an actor (901A) mapped to the “who” portion of the event frame, identify an action (901B) mapped to the “did what” portion of the event frame, identify an object (901C) mapped to the “to whom” portion of the event frame, identify a location (901D) mapped to the “where” portion of the event frame, and identify a timeframe (901E) mapped to the “when” portion of the event frame, [0187]); and
classifying the combined feature set into one or more output classes, and thereby performing the hierarchical classification of data (Computing system 310 may output a rendering of the motion information (1004). For instance, in some examples, animation module 324 may output information about the exemplar animations 380 to a similarity function in order to identify, within a data store of videos (e.g., data store 505), matching or similar videos within the data store, [0194]. User interfaces may be designed such that they enable visualization of explanations generated by the model and allow the user to drill down on decisions, by navigating through a hierarchy or list of exemplar videos to find a video that is relevant (or most relevant) to a search query, [0173]).
Singaraju then discloses classifying the combined feature set into one or more output classes, and thereby performing the hierarchical classification of data (Classification model M4 associated with a first node on the third layer may be trained to classify an input into a group of classes including classes C1-C2 and another group of classes including classes C3-C4. Classification model M5 associated with a second node on the third layer may be trained to classify an input into a group of classes including classes C5-C6 and another group of classes including classes C7-C8, [0064]-[0068]), wherein classifying the combined feature set into one or more output classes is performed using a classifier comprising a plurality of nodes, each said node comprising a neural network (The hierarchical classification model may have a tree structure that includes a plurality of nodes, where each node in the tree structure may be associated with a classification model of any type such as a neural network classifier, [0063]).
It would have been obvious to an ordinary person skilled in the art at the time of the effective filing date of the invention to incorporate the teachings of Singaraju with the teachings of Amer for the purpose of using a hierarchical classification model to classify input datasets based on a combination of features extracted from the datasets and predict outcomes associated with the extracted features.
Regarding claims 2, 10, and 19, Singaraju further discloses classifying the combined feature set into the one or more output classes using a multilevel classifier (The hierarchical classification model may have a tree structure that includes a plurality of nodes, where each node in the tree structure may be associated with a classification model of any type such as a neural network classifier, [0063]).
Regarding claims 4, and 12, Singaraju further discloses the multilevel classifier is trained with training data associated with the one or more output classes (Classification model M4 associated with a first node on the third layer may be trained to classify an input into a group of classes including classes C1-C2 and another group of classes including classes C3-C4. Classification model M5 associated with a second node on the third layer may be trained to classify an input into a group of classes including classes C5-C6 and another group of classes including classes C7-C8, [0064]-[0068]).
Regarding claims 5, and 13, Singaraju further discloses the input data is in the form of one or more modalities, the one or more modalities correspond to an input type (the bot system may receive a message from a Short Message Service (SMS). While discussion herein may refer to communications that the bot system receives as a message, a person of ordinary skill in the art will recognize that the message may be an HTTP post call message, a SMS message, or any other type of communication between two systems, [0038]).
Regarding claims 6, and 14, Singaraju further discloses the multilevel classifier is configured to facilitate taxonomy classification associated with the one or more modalities (The hierarchical classification model may classify each input into one of the classes using two or more classification models at two or more layers of the hierarchical classification model, [0063]. The fourth layer of the tree structure may include eight (2.sup.3) leaf nodes, where the classification model associated with each leaf node may be a binary classification model trained to classify an input into one of two classes, such as a linear classifier (e.g., a logistic regression or naive Bayes classifier), a support vector machine (SVM) classifier, a decision tree classifier, a nearest neighbor classifier, or a neural network classifier, [0068]).
Claims 3, 11, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Amer in view of Singaraju and further in view of Lee et al. (Pub. No. US 2019/0392729, filed on June 19, 2019; hereinafter Lee).
Regarding claims 3, and 11, Lee then discloses the multilevel classifier is a hierarchical multilevel network of interconnected deep neural network models (machine vision algorithms utilize a deep learning network (DLN), for example a convolutional neural network (CNN). Neural networks are computer systems inspired by the human brain. They can be viewed as parallel, densely interconnected computational models that adaptively learn through automatic adjustment of system parameters based on training data, [0099]).
It would have been obvious to an ordinary person skilled in the art at the time of the effective filing date of the invention to incorporate the teachings of Lee with the teachings of Amer, as modified by Singaraju, for the purpose of extracting features associated with input data for classifying the input data by a applying a convolutional neural network onto extracted features of the input data.
Regarding claim 20, Lee further discloses the multilevel classifier is a hierarchical multilevel network of interconnected deep neural network models (machine vision algorithms utilize a deep learning network (DLN), for example a convolutional neural network (CNN). Neural networks are computer systems inspired by the human brain. They can be viewed as parallel, densely interconnected computational models that adaptively learn through automatic adjustment of system parameters based on training data, [0099]).
Singaraju further discloses the multilevel classifier is trained with training data associated with the one or more output classes (The hierarchical classification model may have a tree structure that includes a plurality of nodes, where each node in the tree structure may be associated with a classification model of any type such as a neural network classifier, [0063])
Claims 7-8, and 15-16 are rejected under 35 U.S.C. 103 as being unpatentable over Amer in view of Singaraju and further in view of Liu (Pub. No. US 2019/0220824, published on July 18, 2019).
Regarding claims 7, and 15, Liu then discloses the input data comprises at least one of candidate profiles, resumes, songs, or movies (In step 301, resume data and the optional JOR data is fed to the system. In step 302, the system checks if the resume and JOR data is processed, e.g., presented as structured data with parameters readily to be parsed by the RDTE 203, [0036]).
It would have been obvious to an ordinary person skilled in the art at the time of the effective filing date of the invention to incorporate the teachings of Liu with the teachings of Amer, as modified by Singaraju, for the purpose of utilizing a predictive model to generate matching data regarding a plurality of resume records data relative to the one or more job descriptions using the predictive model.
Regarding claims 8, and 16, Liu further discloses the one or more features comprises at least one of job skills or educational qualification (the model may learn that a software engineer would have a higher level of success in a position of software architect if he or she advances his/her career from “software engineer” to “senior software engineer” in 5 years than another software engineer who takes more than 10 years to achieve the same senior position. His/her resume would produce a very high career path success match score for a specific candidate based on their past resume data, [0050]).
Claim 17 is rejected under 35 U.S.C. 103 as being unpatentable over Amer in view of Singaraju and further in view of Huang et al. (Pub. No. US 2018/0165547, published on June 14, 2018; hereinafter Huang).
Regarding claim 17, Huang then discloses the at least one processing device is configured to classify the data until a terminal class node is reached (classifying until reaching the leaf node of the hierarchical classifier, [0112]-[0117]).
It would have been obvious to an ordinary person skilled in the art at the time of the effective filing date of the invention to incorporate the teachings of Huang with the teachings of Amer, as modified by Singaraju, for the purpose of extracting features of input data based on a convolutional neural network and classifying the input data with respective to multiple output classes.
Related Prior Art
The following references are deemed relevant to the claims:
Ristoski et al. (Pub. No. US 2020/0097602) teaches target ontologies that represent the best match exploiting domain specific word embeddings as features. A top-down, per-node local classifier is utilized wherein a deep convolutional neural network classifier is built for each node (apart from leaves) in the hierarchy: all children nodes are considered as classes for the model, and their corresponding leaves as instances (to train the model). This approach can easily cope with large number of classes in the lower parts of the hierarchy.
Otsuka et al. (Pub. No. US 2006/0140449) teaches a classifier being a neural network having a hierarchical structure composed of the input layer of 92 nodes, a hidden layer of 96 nodes and an output layer of 2 nodes. The values fed to each node of the hidden layer are all added, multiplied by the corresponding allotted weight coefficient, and supplied to the corresponding node of the output layer. In addition, the values fed to each node of the output layer are summed, and finally produced as the value of each of the nodes of the output layer.
Contact Information
Any inquiry concerning this communication or earlier communications from the Examiner should be directed to Son Hoang whose telephone number is (571) 270-1752. The Examiner can normally be reached on Monday – Friday (7:00 AM – 4:00 PM).
If attempts to reach the Examiner by telephone are unsuccessful, the Examiner’s supervisor, Sherief Badawi can be reached on (571) 272-9782. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
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.
/SON T HOANG/ Primary Examiner, Art Unit 2169 September 25, 2026