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 .
Information Disclosure Statement
Acknowledgment is made of the Information Disclosure Statement dated 10/13/2023. All of the cited references have been considered.
Drawings
The drawings have been received on 10/13/2023. These drawings are accepted.
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.
Claims 1-20 rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding Claim 1,
Claim 1 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 1 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“generating a plurality of individual datasets, each of the plurality of individual datasets comprising data of at least one answer to at least one of a plurality of questions of each of a plurality of questionnaires;”
“generating a first batch of individual datasets, the first batch of individual datasets comprising one or more of the plurality of individual datasets;”
“encoding the data of the first batch of individual datasets [with an autoencoder.]”
As drafted, under their broadest reasonable interpretation, cover concepts performed in human mind (including an observation, evaluation, judgement, or opinion, e.g., generating, encoding). The above limitations in the context of this claim encompass, inter alia, generating a plurality of individual datasets, encoding the data (corresponding to mental processes which can be done mentally or by pen and paper). Examiner is interpreting generating as creating answers to questions.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application.
The limitations:
“[encoding the data of the first batch of individual datasets] with an autoencoder.”
As drafted, are additional elements that amount to no more than mere instructions to apply the exception for the abstract ideas. See MPEP 2106.05(f). Specifically, they amount to mere instructions to apply the exception using an autoencoder (e.g., by using these elements as tools).
The limitations:
“inputting the first batch of individual datasets into the artificial intelligence model; and”
As drafted, amount to insignificant extra-solution activities, which do not integrate a judicial exception into a practical application. For example, the additional elements of "inputting the first batch of individual datasets" amount to mere data gathering and data storage, respectively, which are insignificant extra-solution activities that do not integrate a judicial exception into a practical application. See MPEP 2106.05(g).
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
The limitations:
“[encoding the data of the first batch of individual datasets] with an autoencoder.”
As drafted, are additional elements that amount to no more than mere instructions to apply the exception for the abstract ideas. See MPEP 2106.05(f). Specifically, they amount to mere instructions to apply the exception using an autoencoder (e.g., by using these elements as tools).
As discussed above with respect to integration of the abstract idea into a practical application, all of the additional elements are insignificant extra-solution activities or mere instructions to apply an exception. (i.e., the additional element describes a unit for applying the abstract ideas). Insignificant extra-solution activities and mere instructions to apply an exception cannot provide an inventive concept. Moreover, receiving, communicating, and storing data are insignificant extra-solution activities that are well-understood, routine, and conventional. See MPEP 2106.05(d)(II) ("The courts have recognized the following computer functions as well-understood, routine, and conventional functions ... i. Receiving or transmitting data over a network") (citing OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015)).
The claim is not patent eligible.
Regarding Claim 2,
Claim 2 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 2 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“generating the plurality of questionnaires, each of the plurality of questionnaires comprising one or more questions.”
As drafted, under their broadest reasonable interpretation, cover concepts performed in human mind (including an observation, evaluation, judgement, or opinion, e.g., generating). The above limitations in the context of this claim encompass, inter alia, generating the plurality of questionnaires (corresponding to mental processes which can be done mentally or by pen and paper).
Step 2A Prong Two Analysis: Please see the corresponding analysis of Claim 1.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 3,
Claim 3 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 3 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“generating a second batch of individual datasets, the second batch of individual datasets comprising a different combination of one or more of the plurality of individual datasets than the first batch of individual datasets.”
As drafted, under their broadest reasonable interpretation, cover concepts performed in human mind (including an observation, evaluation, judgement, or opinion, e.g., generating). The above limitations in the context of this claim encompass, inter alia, generating a second batch of individual datasets (corresponding to mental processes which can be done mentally or by pen and paper). Examiner is interpreting generating as creating answers to questions.
Step 2A Prong Two Analysis: Please see the corresponding analysis of Claim 1.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 4,
Claim 4 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 4 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“wherein the first batch of individual datasets comprises a matrix of one or more answers to the plurality of questionnaires.”
As drafted, under their broadest reasonable interpretation, cover concepts performed in human mind (including an observation, evaluation, judgement, or opinion, e.g., generating). The above limitations in the context of this claim encompass, inter alia, generating a first batch of individual datasets (corresponding to mental processes which can be done mentally or by pen and paper). Examiner is interpreting generating as creating answers to questions.
Step 2A Prong Two Analysis: Please see the corresponding analysis of Claim 1.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 5,
Claim 5 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 5 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“wherein the matrix is sized to such that one answer to each of the plurality of questions of the plurality of questionnaires is included in the matrix.”
As drafted, under their broadest reasonable interpretation, cover concepts performed in human mind (including an observation, evaluation, judgement, or opinion, e.g., generating). The above limitations in the context of this claim encompass, inter alia, generating a first batch of individual datasets (corresponding to mental processes which can be done mentally or by pen and paper). Examiner is interpreting generating as creating answers to questions.
Step 2A Prong Two Analysis: Please see the corresponding analysis of Claim 1.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 6,
Claim 6 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 6 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“wherein the matrix uses zero masking when the first batch of individual datasets comprises less than one answer to each of the plurality of questions of the plurality of questionnaires.”
As drafted, under their broadest reasonable interpretation, cover concepts performed in human mind (including an observation, evaluation, judgement, or opinion, e.g., generating). The above limitations in the context of this claim encompass, inter alia, generating a first batch of individual datasets (corresponding to mental processes which can be done mentally or by pen and paper). Examiner is interpreting generating as creating answers to questions.
Step 2A Prong Two Analysis: Please see the corresponding analysis of Claim 1.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 7,
Claim 7 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 7 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“wherein a plurality of users provides answers to the plurality of questions of each of the plurality of questionnaires.”
As drafted, under their broadest reasonable interpretation, cover concepts of organizing human activity (including fundamental economic principles, commercial or legal interactions, or managing personal behavior or relationships or interactions between people, e.g., wherein a plurality of users). The above limitations in the context of this claim encompass, inter alia, wherein a plurality of users provides answers to the plurality of questions (corresponding to organizing human activity).
Step 2A Prong Two Analysis: Please see the corresponding analysis of Claim 1.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 8,
Claim 8 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 8 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“wherein the plurality of questionnaires comprises at least one of a pHEV questionnaire, a gambling questionnaire, an AoT questionnaire, a demographic questionnaire, a pro-social questionnaire, or a big five questionnaire.”
As drafted, under their broadest reasonable interpretation, cover concepts performed in human mind (including an observation, evaluation, judgement, or opinion, e.g., generating). The above limitations in the context of this claim encompass, inter alia, generating a plurality of individual datasets, (corresponding to mental processes which can be done mentally or by pen and paper). Examiner is interpreting generating as creating answers to questions.
Step 2A Prong Two Analysis: Please see the corresponding analysis of Claim 1.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 9,
Claim 9 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 9 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“wherein the artificial intelligence model is measured with a loss function.”
As drafted, under their broadest reasonable interpretation, cover mathematical concepts (including mathematical relationships, mathematical formulas or equations, or mathematical calculations, e.g., loss function). The above limitations in the context of this claim encompass, inter alia, measuring the artificial intelligence model with a loss function (corresponding to mathematical concepts). See paragraphs [0024] and [0064] of the Specification to see mathematical concepts and formulas related to the loss function.
Step 2A Prong Two Analysis: Please see the corresponding analysis of Claim 1.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 10,
Claim 10 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 10 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“wherein the loss function comprises the sum of a choice prediction loss and a reconstruction loss, and wherein the choice prediction loss and the reconstruction loss are the mean squared error between a true target and a predicted output of the artificial intelligence model.”
As drafted, under their broadest reasonable interpretation, cover mathematical concepts (including mathematical relationships, mathematical formulas or equations, or mathematical calculations, e.g., loss function). The above limitations in the context of this claim encompass, inter alia, measuring the artificial intelligence model with a loss function (corresponding to mathematical concepts). See paragraphs [0024] and [0064] of the Specification to see mathematical concepts and formulas related to the loss function.
Step 2A Prong Two Analysis: Please see the corresponding analysis of Claim 1.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 11,
Claim 11 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 11 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“wherein the plurality of questionnaires are embedded within a joint embedding space.”
As drafted, under their broadest reasonable interpretation, cover concepts performed in human mind (including an observation, evaluation, judgement, or opinion, e.g., embedding). The above limitations in the context of this claim encompass, inter alia, embedding questionnaires within a joint embedding space (corresponding to mental processes which can be done mentally or by pen and paper).
Step 2A Prong Two Analysis: Please see the corresponding analysis of Claim 1.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claim 12,
Claim 12 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 12 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: Please see the corresponding analysis of Claim 1.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
The limitations:
“wherein the artificial intelligence model comprises a plurality of transformer blocks.”
The following additional elements are directed to an artificial intelligence model comprising a plurality of transformer blocks at a high level of generality. Huang et al. (TabTransformer: Tabular Data Modeling Using Contextual Embeddings), hereinafter Huang discloses on page 2, “self-attention based Transformers (Vaswani et al. 2017) have become a standard component of NLP models to achieve state-of-the-art performance. The effectiveness and interpretability of contextual embeddings generated by Transformers have been also well studied (Co-enen et al. 2019; Brunner et al. 2019).” Huang has recognized self-attention based transformers in artificial intelligence models as well-understood, routine, and conventional activity previously known in the industry [see MPEP 2106.05(d)].
The claim is not patent eligible.
Regarding Claim 13,
Claim 13 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 13 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: Please see the corresponding analysis of Claim 1.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
The limitations:
“testing the artificial intelligence model with a testing dataset, the testing dataset comprising fewer answers to the plurality of questions of each of the plurality of questionnaires than the first batch of individual datasets.”
The following additional elements are directed to testing the artificial intelligence model with a testing dataset at a high level of generality. Tan et al. (Acritical look at the current train/test split in machine learning), hereinafter Tan, discloses in Abstract, “The randomized or cross-validated split of training and testing sets has been adopted as the gold standard of machine learning for decades. The establishment of these split protocols are based on two assumptions: (i)-fixing the dataset to be eternally static so we could evaluate different machine learning algorithms or models;” Tan has recognized using a testing set to evaluate machine learning models as a well-understood, routine, and conventional activity previously known in the industry [see MPEP 2106.05(d)].
Regarding Claim 14,
Claim 14 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 14 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: Please see the corresponding analysis of Claim 1.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
The limitations:
“wherein the testing dataset comprises one or more answers to one of the plurality of individual datasets, and the first batch of individual datasets comprises one or more answers to a plurality of the plurality of individual datasets.”
The following additional elements are directed to testing the artificial intelligence model with a testing dataset at a high level of generality. Tan et al. (Acritical look at the current train/test split in machine learning), hereinafter Tan, discloses in Abstract, “The randomized or cross-validated split of training and testing sets has been adopted as the gold standard of machine learning for decades. The establishment of these split protocols are based on two assumptions: (i)-fixing the dataset to be eternally static so we could evaluate different machine learning algorithms or models;” Tan has recognized using a testing set to evaluate machine learning models as a well-understood, routine, and conventional activity previously known in the industry [see MPEP 2106.05(d)].
Regarding Claim 15,
Claim 15 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 15 is directed to a method, i.e., a process, one of the statutory categories.
The limitations:
“determining if the artificial intelligence model exceeds a predetermined performance threshold; and”
“if the artificial intelligence model does not exceed the predetermined performance threshold:
generating a second batch of individual datasets;”
“encoding the data of the second batch of individual datasets [with the autoencoder.]”
As drafted, under their broadest reasonable interpretation, cover concepts performed in human mind (including an observation, evaluation, judgement, or opinion, e.g., determining, generating, encoding). The above limitations in the context of this claim encompass, inter alia, determining if the artificial intelligence model exceeds a predetermined performance threshold, generating a second batch of individual datasets, encoding the data (corresponding to mental processes which can be done mentally or by pen and paper).
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application.
The limitations:
“[encoding the data of the second batch of individual datasets] with the autoencoder.”
As drafted, are additional elements that amount to no more than mere instructions to apply the exception for the abstract ideas. See MPEP 2106.05(f). Specifically, they amount to mere instructions to apply the exception using an autoencoder (e.g., by using these elements as tools).
The limitations:
“inputting the second batch of individual datasets into the artificial intelligence model; and”
As drafted, amount to insignificant extra-solution activities, which do not integrate a judicial exception into a practical application. For example, the additional elements of "inputting the first batch of individual datasets" amount to mere data gathering and data storage, respectively, which are insignificant extra-solution activities that do not integrate a judicial exception into a practical application. See MPEP 2106.05(g).
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
The limitations:
“[encoding the data of the first batch of individual datasets] with an autoencoder.”
As drafted, are additional elements that amount to no more than mere instructions to apply the exception for the abstract ideas. See MPEP 2106.05(f). Specifically, they amount to mere instructions to apply the exception using an autoencoder (e.g., by using these elements as tools).
As discussed above with respect to integration of the abstract idea into a practical application, all of the additional elements are insignificant extra-solution activities or mere instructions to apply an exception. (i.e., the additional element describes a unit for applying the abstract ideas). Insignificant extra-solution activities and mere instructions to apply an exception cannot provide an inventive concept. Moreover, receiving, communicating, and storing data are insignificant extra-solution activities that are well-understood, routine, and conventional. See MPEP 2106.05(d)(II) ("The courts have recognized the following computer functions as well-understood, routine, and conventional functions ... i. Receiving or transmitting data over a network") (citing OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015)).
The claim is not patent eligible.
Regarding Claim 16,
Claim 16 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 16 is directed to a method, i.e., a process, one of the statutory categories.
The limitations:
“[the trained artificial intelligence model] predicting an answer to at least one question from at least one of a plurality of questionnaires, each of the plurality of questionnaires being different than the baseline questionnaire.”
As drafted, under their broadest reasonable interpretation, cover concepts performed in human mind (including an observation, evaluation, judgement, or opinion, e.g., predicting). The above limitations in the context of this claim encompass, inter alia, predicting an answer (corresponding to mental processes which can be done mentally or by pen and paper).
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application.
The limitations:
“the trained artificial intelligence model comprising a self-attention layer, the self-attention layer creating a latent vector of the at least one answer to the at least one question from the baseline questionnaire;”
“the trained artificial intelligence model [predicting an answer to at least one question from at least one of a plurality of questionnaires, each of the plurality of questionnaires being different than the baseline questionnaire.]”
As drafted, are additional elements that amount to no more than mere instructions to apply the exception for the abstract ideas. See MPEP 2106.05(f). Specifically, they amount to mere instructions to apply the exception using an artificial intelligence based model (e.g., by using these elements as tools).
The limitations:
“feeding the latent vector through a decoder of the trained artificial intelligence model; and”
As drafted, amount to insignificant extra-solution activities, which do not integrate a judicial exception into a practical application. For example, the additional elements of "feeding the latent vector through a decoder" amount to mere data gathering and data storage, respectively, which are insignificant extra-solution activities that do not integrate a judicial exception into a practical application. See MPEP 2106.05(g).
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 with respect to integration of the abstract idea into a practical application, all of the additional elements are insignificant extra-solution activities or mere instructions to apply an exception. (i.e., the additional element describes a unit for applying the abstract ideas). Insignificant extra-solution activities and mere instructions to apply an exception cannot provide an inventive concept. Moreover, receiving, communicating, and storing data are insignificant extra-solution activities that are well-understood, routine, and conventional. See MPEP 2106.05(d)(II) ("The courts have recognized the following computer functions as well-understood, routine, and conventional functions ... i. Receiving or transmitting data over a network") (citing OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015)).
The claim is not patent eligible.
Regarding Claim 17,
Claim 17 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 17 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: Please see the corresponding analysis of Claim 1.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application.
The limitations:
“the baseline questionnaire comprising a plurality of questions; and”
“the trained artificial intelligence model receiving a plurality of answers to the plurality of questions of the baseline questionnaire.”
As drafted, amount to insignificant extra-solution activities, which do not integrate a judicial exception into a practical application. For example, the additional elements of "receiving a plurality of answers to the plurality of questions of the baseline questionnaire." amount to mere data gathering and data storage, respectively, which are insignificant extra-solution activities that do not integrate a judicial exception into a practical application. See MPEP 2106.05(g).
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 with respect to integration of the abstract idea into a practical application, all of the additional elements are insignificant extra-solution activities or mere instructions to apply an exception. (i.e., the additional element describes a unit for applying the abstract ideas). Insignificant extra-solution activities and mere instructions to apply an exception cannot provide an inventive concept. Moreover, receiving, communicating, and storing data are insignificant extra-solution activities that are well-understood, routine, and conventional. See MPEP 2106.05(d)(II) ("The courts have recognized the following computer functions as well-understood, routine, and conventional functions ... i. Receiving or transmitting data over a network") (citing OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015)).
The claim is not patent eligible.
Regarding Claim 18,
Claim 18 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 18 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: Please see the corresponding analysis of Claim 1.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application.
The limitations:
“wherein the baseline questionnaire comprises at least one of: a pHEV questionnaire, a gambling questionnaire, an AoT questionnaire, a demographic questionnaire, a pro-social questionnaire, or a big five questionnaire.”
As drafted, amount to insignificant extra-solution activities, which do not integrate a judicial exception into a practical application. For example, the additional elements of "receiving a plurality of answers to the plurality of questions of the baseline questionnaire." amount to mere data gathering and data storage, respectively, which are insignificant extra-solution activities that do not integrate a judicial exception into a practical application. See MPEP 2106.05(g).
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 with respect to integration of the abstract idea into a practical application, all of the additional elements are insignificant extra-solution activities or mere instructions to apply an exception. (i.e., the additional element describes a unit for applying the abstract ideas). Insignificant extra-solution activities and mere instructions to apply an exception cannot provide an inventive concept. Moreover, receiving, communicating, and storing data are insignificant extra-solution activities that are well-understood, routine, and conventional. See MPEP 2106.05(d)(II) ("The courts have recognized the following computer functions as well-understood, routine, and conventional functions ... i. Receiving or transmitting data over a network") (citing OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015)).
The claim is not patent eligible.
Regarding Claim 19,
Claim 19 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 19 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: Please see the corresponding analysis of Claim 1.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application.
The limitations:
“wherein the plurality of questionnaires comprises at least one of: a pHEV questionnaire, a gambling questionnaire, an AoT questionnaire, a demographic questionnaire, a pro-social questionnaire, or a big five questionnaire.”
As drafted, amount to insignificant extra-solution activities, which do not integrate a judicial exception into a practical application. For example, the additional elements of "receiving a plurality of answers to the plurality of questions of the baseline questionnaire." amount to mere data gathering and data storage, respectively, which are insignificant extra-solution activities that do not integrate a judicial exception into a practical application. See MPEP 2106.05(g).
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 with respect to integration of the abstract idea into a practical application, all of the additional elements are insignificant extra-solution activities or mere instructions to apply an exception. (i.e., the additional element describes a unit for applying the abstract ideas). Insignificant extra-solution activities and mere instructions to apply an exception cannot provide an inventive concept. Moreover, receiving, communicating, and storing data are insignificant extra-solution activities that are well-understood, routine, and conventional. See MPEP 2106.05(d)(II) ("The courts have recognized the following computer functions as well-understood, routine, and conventional functions ... i. Receiving or transmitting data over a network") (citing OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015)).
The claim is not patent eligible.
Regarding Claim 20,
Claim 20 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 20 is directed to a method, i.e., a process, one of the statutory categories.
Step 2A Prong One Analysis: Please see the corresponding analysis of Claim 1.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application.
The limitations:
“further comprising feeding the latent vector through a multi-layer perceptron of the trained artificial intelligence model.”
As drafted, amount to insignificant extra-solution activities, which do not integrate a judicial exception into a practical application. For example, the additional elements of " feeding the latent vector through a multi-layer perceptron" amount to mere data gathering and data storage, respectively, which are insignificant extra-solution activities that do not integrate a judicial exception into a practical application. See MPEP 2106.05(g).
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 with respect to integration of the abstract idea into a practical application, all of the additional elements are insignificant extra-solution activities or mere instructions to apply an exception. (i.e., the additional element describes a unit for applying the abstract ideas). Insignificant extra-solution activities and mere instructions to apply an exception cannot provide an inventive concept. Moreover, receiving, communicating, and storing data are insignificant extra-solution activities that are well-understood, routine, and conventional. See MPEP 2106.05(d)(II) ("The courts have recognized the following computer functions as well-understood, routine, and conventional functions ... i. Receiving or transmitting data over a network") (citing OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015)).
The claim is not patent eligible.
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, 3, 4, 5, 7, 8, 9, 11, 13, 14, 16, 17, 18, 19 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Purpura et al. (Identifying Faked Responses in Questionnaires with Self-Attention-Based Autoencoders); hereinafter Purpura in view of Kim et al. (AI-Augmented Surveys: Leveraging Large Language Models for Opinion Prediction in Nationally Representative Surveys); hereinafter Kim
Claim 1 is rejected over Purpura and Kim.
Regarding claim 1, Purpura teaches a method for training an artificial intelligence model, the method comprising: (Purpura [page 5, Training Strategy]: “The training strategy we employ is similar to the masked language model (MLM) paradigm described in the bidirectional encoder representations from transformers (BERT) paper [4]. We feed a sequence of honest responses to our model, where we mask one or more of them randomly with a special mask token ([M]). The model is then trained to predict the values of the responses r1, . . . , rn that were masked in the input sequence,”)
generating a plurality of individual datasets, each of the plurality of individual datasets comprising data of at least one answer to at least one of a plurality of questions [of each of a plurality of questionnaires;] (Purpura [page 6, 4. Materials & Methods]: “We collected data from a total of 694 participants who responded online to the personality
questionnaire described below. The number of participants that took part in each of the three data collection experiments are reported in Table 1. The test subjects in each dataset are distinct and responded to our questionnaire twice, once honestly and once faking their responses.”
generating a first batch of individual datasets, the first batch of individual datasets comprising one or more of the plurality of individual datasets; (Purpura [page 8]: “The hyperparameters of our model are optimized separately on a random validation set—sampled each time from the training data—considering the following grid of points: number of items to mask during training n ∈ {1, 3, 5}, batch size b ∈ {16, 64, 128},”)
inputting the first batch of individual datasets into the artificial intelligence model; and (Purpura [page 4]: “Specifically, when the model receives a new input sequence i1, . . . , in (a participant’s sequence of responses to the n = 10 test items), the corresponding values in the embedding matrix—that are trained as parameters of our model—are rescaled depending on the value of each item in the sequence.”)
encoding the data of the first batch of individual datasets with an autoencoder. (Purpura [page 2]: “we propose to use an unsupervised deep learning approach based on an autoencoder. An autoencoder [3] is an unsupervised deep learning model that learns to reconstruct its input from a latent hidden representation. Given a set of answers to a questionnaire, we train the proposed model to reconstruct masked responses [4] in the input based on the remaining unmasked ones.”)
Purpura does not appear to explicitly teach of each of a plurality of questionnaires;
a plurality of questions of each of a plurality of questionnaires;
However, Kim teaches of each of a plurality of questionnaires;
a plurality of questions of each of a plurality of questionnaires; (Kim [page 19]: “The GSS dataset provides comprehensive information about the demographic characteristics, political and ideological beliefs, cultural tastes, personal morality, and religiosity of people in the United States. We use 68,846 individuals’ responses to 3,110 questions collected for 33 repeated cross-sectional data between 1972 and 2021 for fine-tuning the LLMs.”)
It would have been obvious before the effective filing date to combine self-attention based autoencoders of Purpura with the opinion prediction of questionnaires of Kim to improve survey quality (Kim, page 14). Kim and Purpura are analogous art because they both concern predicting answers to questionnaires.
Claim 2 is rejected over Purpura and Kim with the incorporation of claim 1.
Regarding claim 2, Purpura teaches generating the [plurality of] questionnaires, each of the plurality of questionnaires comprising one or more questions. (Purpura [page 7]: “This questionnaire is a short version (developed forWeb applications) aimed at profiling the personality of the respondent according to the Big Five model [5], mapping the subject responses into five orthogonal dimensions (extroversion, agreeableness, conscientiousness, emotional stability, and openness). We asked the participants to take our test twice. First, we asked them to answer honestly to all questions and later to take the test pretending to be in one of the following situations when a person would be likely to lie on his/her personality traits: (i) during a psychological assessment to earn the custody of his/her children in a divorce (CC), (ii) during a job interview, to be hired as a salesperson (JIS), and (iii) during a job interview, to be hired by a humanitarian organization (JIHO).”; Note: See Table 1 to see three datasets of questionnaires”)
Purpura does not appear to explicitly teach generating the plurality of questionnaires, each of the plurality of questionnaires comprising one or more questions.
However, Kim teaches generating the plurality of questionnaires, each of the plurality of questionnaires comprising one or more questions. (Kim [page 19]: “The GSS dataset provides comprehensive information about the demographic characteristics, political and ideological beliefs, cultural tastes, personal morality, and religiosity of people in the United States. We use 68,846 individuals’ responses to 3,110 questions collected for 33 repeated cross-sectional data between 1972 and 2021 for fine-tuning the LLMs.”)
It would have been obvious before the effective filing date to combine self-attention based autoencoders of Purpura with the opinion prediction of questionnaires of Kim to improve survey quality (Kim, page 14). Kim and Purpura are analogous art because they both concern predicting answers to questionnaires.
Claim 3 is rejected over Purpura and Kim with the incorporation of claim 1.
Regarding claim 3, Purpura teaches generating a second batch of individual datasets, the second batch of individual datasets comprising a different combination of one or more of the plurality of individual datasets than the first batch of individual datasets. (Purpura [page 8]: “to compute the performance of the proposed approach on the CC dataset, we train our model performing 10-fold cross-validation—considering 10 different training–validation splits—only on the honest responses in the JIS and JIHO datasets. For each fold, we evaluate the performance of the model on the same test dataset.”)
Claim 4 is rejected over Purpura and Kim with the incorporation of claim 1.
Regarding claim 4, Purpura does not appear to explicitly teach wherein the first batch of individual datasets comprises a matrix of one or more answers to the plurality of questionnaires.
However, Kim teaches wherein the first batch of individual datasets comprises a matrix of one or more answers to the plurality of questionnaires. (Kim [page 19, Model training]: “we first construct a matrix of survey responses where each row represents an individual, each column represents a survey question, and each element indicates the individual’s response to the survey question.”)
It would have been obvious before the effective filing date to combine self-attention based autoencoders of Purpura with the opinion prediction of questionnaires of Kim to improve survey quality (Kim, page 14). Kim and Purpura are analogous art because they both concern predicting answers to questionnaires.
Claim 5 is rejected over Purpura and Kim with the incorporation of claim 1.
Regarding claim 5, Purpura does not appear to explicitly teach wherein the matrix is sized to such that one answer to each of the plurality of questions of the plurality of questionnaires is included in the matrix.
However, Kim teaches wherein the matrix is sized to such that one answer to each of the plurality of questions of the plurality of questionnaires is included in the matrix. (Kim [page 19, Model training]: “we first construct a matrix of survey responses where each row represents an individual, each column represents a survey question, and each element indicates the individual’s response to the survey question.”)
It would have been obvious before the effective filing date to combine self-attention based autoencoders of Purpura with the opinion prediction of questionnaires of Kim to improve survey quality (Kim, page 14). Kim and Purpura are analogous art because they both concern predicting answers to questionnaires.
Claim 7 is rejected over Purpura and Kim with the incorporation of claim 1.
Regarding claim 7, Purpura teaches wherein a plurality of users provides answers to the plurality of questions of each of the plurality of questionnaires. (Purpura [page 7]: “We asked the participants to take our test twice. First, we asked them to answer honestly to all questions and later to take the test pretending to be in one of the following situations when a person would be likely to lie on his/her personality traits: (i) during a psychological assessment to earn the custody of his/her children in a divorce (CC), (ii) during a job interview, to be hired as a salesperson (JIS), and (iii) during a job interview, to be hired by a humanitarian organization (JIHO).”)
Claim 8 is rejected over Purpura and Kim with the incorporation of claim 1.
Regarding claim 8, Purpura teaches wherein the plurality of questionnaires comprises at least one of a pHEV questionnaire, a gambling questionnaire, an AoT questionnaire, a demographic questionnaire, a pro-social questionnaire, or a big five questionnaire. (Purpura [page 7]: “This questionnaire is a short version (developed forWeb applications) aimed at profiling the personality of the respondent according to the Big Five model [5], mapping the subject responses into five orthogonal dimensions (extroversion, agreeableness, conscientiousness, emotional stability, and openness).;”; Note: See Table 1 to see three datasets of questionnaires”)
Claim 9 is rejected over Purpura and Kim with the incorporation of claim 1.
Regarding claim 9, Purpura teaches wherein the artificial intelligence model is measured with a loss function. (Purpura [page 5]: “The model is then trained to predict the values of the responses r1, . . . , rn that were masked in the input sequence, minimizing the mean absolute error (MAE) between the predicted and the true values.”)
Claim 11 is rejected over Purpura and Kim with the incorporation of claim 1.
Regarding claim 11, Purpura does not appear to explicitly teach wherein the plurality of questionnaires are embedded within a joint embedding space.
However, Kim teaches wherein the plurality of questionnaires are embedded within a joint embedding space. (Purpura [page 18]: “We begin by concatenating the semantic embedding of the survey question (s), the individual’s belief embedding (b), and the temporal embedding (p) into a single vector, denoted as x0. The (l+1)th cross layer can then be defined as follows:”; [page 4]: “these neural embeddings represent similarities in the meanings of survey questions, individual beliefs, and temporal contexts in high-dimensional vector spaces. Then, our models use these latent features to predict the most plausible answer to a specific question for each individual at a given moment … response patterns being closely located in the embedding space.”)
It would have been obvious before the effective filing date to combine self-attention based autoencoders of Purpura with the embeddings of Kim to improve survey quality (Kim, page 14). Kim and Purpura are analogous art because they both concern predicting answers to questionnaires.
Claim 13 is rejected over Purpura and Kim with the incorporation of claim 1.
Regarding claim 13, Purpura teaches testing the artificial intelligence model with a testing dataset, the testing dataset comprising fewer answers to the plurality of questions of each of the plurality of questionnaires than the first batch of individual datasets. (Purpura [page 8]: “we train the proposed model on the honest data from two of the three datasets with 10-fold cross validation and early stopping with patience 20—considering the MAE on the reconstruction in the validation set of each fold—and report the average performance the model achieved on the remaining dataset. For example, to compute the performance of the proposed approach on the CC dataset, we train our model performing 10-fold cross-validation—considering 10 different training–validation splits—only on the honest responses in the JIS and JIHO datasets. For each fold, we evaluate the performance of the model on the same test dataset.”)
Claim 14 is rejected over Purpura and Kim with the incorporation of claim 1.
Regarding claim 14, Purpura teaches wherein the testing dataset comprises one or more answers to one of the plurality of individual datasets, and the first batch of individual datasets comprises one or more answers to a plurality of the plurality of individual datasets. (Purpura [page 8]: “we train the proposed model on the honest data from two of the three datasets with 10-fold cross validation and early stopping with patience 20—considering the MAE on the reconstruction in the validation set of each fold—and report the average performance the model achieved on the remaining dataset. For example, to compute the performance of the proposed approach on the CC dataset, we train our model performing 10-fold cross validation—considering 10 different training–validation splits—only on the honest responses in the JIS and JIHO datasets.”)
Claim 16 is rejected over Purpura and Kim.
Regarding claim 16, Purpura teaches a method for predicting a task output with a trained artificial intelligence model, the method comprising: (Purpura [page 5, Training Strategy]: “The training strategy we employ is similar to the masked language model (MLM) paradigm described in the bidirectional encoder representations from transformers (BERT) paper [4]. We feed a sequence of honest responses to our model, where we mask one or more of them randomly with a special mask token ([M]). The model is then trained to predict the values of the responses r1, . . . , rn that were masked in the input sequence,”)
the trained artificial intelligence model receiving at least one answer to at least one question from a baseline questionnaire; (Purpura [page 4]: “Specifically, when the model receives a new input sequence i1, . . . , in (a participant’s sequence of responses to the n = 10 test items), the corresponding values in the embedding matrix—that are trained as parameters of our model—are rescaled depending on the value of each item in the sequence.”)
the trained artificial intelligence model comprising a self-attention layer, the self-attention layer creating a latent vector of the at least one answer to the at least one question from the baseline questionnaire; (Purpura [page 4]: “this process allows us to consider the responses as words in a sentence and to apply the same SA layer—depicted in Figure 2—as proposed in the transformer model [20] to rescale the input data leveraging on local attention patterns. The SA layer we employ performs the following well-known self-attention formulation [20] on the items of the the input sequence”; and [page 3]: “An autoencoder is an unsupervised—sometimes also referred as a self-supervised—model which learns to reconstruct its input from a lower-dimensional latent representation. Here, we employ an enhanced version of this model relying on self-attention (SA) [20] to flag faked responses in a questionnaire.”)
feeding the latent vector through a decoder of the trained artificial intelligence model; and (Purpura [page 3]: “An autoencoder is an unsupervised—sometimes also referred as a self-supervised—model which learns to reconstruct its input from a lower-dimensional latent representation. Here, we employ an enhanced version of this model relying on self-attention (SA) [20] to flag faked responses in a questionnaire.”; and [page 4]: “Finally, we employ a feedforward output layer with a sigmoid activation function to estimate the masked input values.”)
Purpura does not appear to explicitly teach the trained artificial intelligence model predicting an answer to at least one question from at least one of a plurality of questionnaires, each of the plurality of questionnaires being different than the baseline questionnaire.
However, Kim teaches the trained artificial intelligence model predicting an answer to at least one question from at least one of a plurality of questionnaires, each of the plurality of questionnaires being different than the baseline questionnaire. (Kim [page 3]: “We use the term “zero-shot prediction” to refer to predicting responses to a question without any prior survey responses about the question in the training data.”; [page 13]: “pollsters can disseminate 20 questions among 100 participants, each answering 10 questions, and employ the model to infer individual responses to the remaining 10 questions that were not directly answered.”; and [page 19]: “The GSS dataset provides comprehensive information about the demographic characteristics, political and ideological beliefs, cultural tastes, personal morality, and religiosity of people in the United States. We use 68,846 individuals’ responses to 3,110 questions”; Note: Kim teaches the plurality of questionnaires.)
It would have been obvious before the effective filing date to combine self-attention based autoencoders of Purpura with the opinion prediction of questionnaires of Kim to improve survey quality (Kim, page 14). Kim and Purpura are analogous art because they both concern predicting answers to questionnaires.
Claim 17 is rejected over Purpura and Kim with the incorporation of claim 16.
Regarding claim 17, Purpura teaches the baseline questionnaire comprising a plurality of questions; and (Purpura [page 7]: “This questionnaire is a short version (developed forWeb applications) aimed at profiling the personality of the respondent according to the Big Five model [5], mapping the subject responses into five orthogonal dimensions (extroversion, agreeableness, conscientiousness, emotional stability, and openness). We asked the participants to take our test twice. First, we asked them to answer honestly to all questions and later to take the test pretending to be in one of the following situations when a person would be likely to lie on his/her personality traits: (i) during a psychological assessment to earn the custody of his/her children in a divorce (CC), (ii) during a job interview, to be hired as a salesperson (JIS), and (iii) during a job interview, to be hired by a humanitarian organization (JIHO).”; Note: See Table 1 to see three datasets of questionnaires”)
the trained artificial intelligence model receiving a plurality of answers to the plurality of questions of the baseline questionnaire. (Purpura [page 4]: “Specifically, when the model receives a new input sequence i1, . . . , in (a participant’s sequence of responses to the n = 10 test items), the corresponding values in the embedding matrix—that are trained as parameters of our model—are rescaled depending on the value of each item in the sequence.”)
Claim 18 is rejected over Purpura and Kim with the incorporation of claim 16.
Regarding claim 18, Purpura teaches wherein the baseline questionnaire comprises at least one of: a pHEV questionnaire, a gambling questionnaire, an AoT questionnaire, a demographic questionnaire, a pro-social questionnaire, or a big five questionnaire. (Purpura [page 7]: “This questionnaire is a short version (developed forWeb applications) aimed at profiling the personality of the respondent according to the Big Five model [5], mapping the subject responses into five orthogonal dimensions (extroversion, agreeableness, conscientiousness, emotional stability, and openness). We asked the participants to take our test twice. First, we asked them to answer honestly to all questions and later to take the test pretending to be in one of the following situations when a person would be likely to lie on his/her personality traits: (i) during a psychological assessment to earn the custody of his/her children in a divorce (CC), (ii) during a job interview, to be hired as a salesperson (JIS), and (iii) during a job interview, to be hired by a humanitarian organization (JIHO).”; Note: See Table 1 to see three datasets of questionnaires”)
Claim 19 is rejected over Purpura and Kim with the incorporation of claim 16.
Regarding claim 19, Purpura teaches wherein the plurality of questionnaires comprises at least one of: a pHEV questionnaire, a gambling questionnaire, an AoT questionnaire, a demographic questionnaire, a pro-social questionnaire, or a big five questionnaire. (Purpura [page 7]: “This questionnaire is a short version (developed forWeb applications) aimed at profiling the personality of the respondent according to the Big Five model [5], mapping the subject responses into five orthogonal dimensions (extroversion, agreeableness, conscientiousness, emotional stability, and openness).;”; Note: See Table 1 to see three datasets of questionnaires”)
Claim 20 is rejected over Purpura and Kim with the incorporation of claim 16.
Regarding claim 20, Purpura teaches further comprising feeding the latent vector [through a multi-layer perceptron] of the trained artificial intelligence model. (Purpura [page 4]: “apply the same SA layer—depicted in Figure 2—as proposed in the transformer model [20] to rescale the input data leveraging on local attention patterns.”; Note: This is the latent vector; and “Finally, we employ a feedforward output layer with a sigmoid activation function to estimate the masked input values.”)
Purpura does not appear to explicitly teach further comprising [feeding the latent vector through] a multi-layer perceptron of the trained artificial intelligence model.
However, Kim teaches further comprising [feeding the latent vector through] a multi-layer perceptron of the trained artificial intelligence model. (Kim [page 19, Model training]: “three feed-forward dense layers of size 150,”)
It would have been obvious before the effective filing date to combine self-attention based autoencoders of Purpura with the multi-layer of Kim to improve survey quality (Kim, page 14). Kim and Purpura are analogous art because they both concern predicting answers to questionnaires.
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Purpura and Kim and in further view of Zhang et al. (US 20210358577 A1); hereinafter Zhang
Claim 6 is rejected over Purpura, Kim and Zhang with the incorporation of claim 1.
Regarding claim 6, Purpura does not appear to explicitly teach wherein the matrix uses zero masking when the first batch of individual datasets comprises less than one answer to each of the plurality of questions of the plurality of questionnaires.
However, Zhang teaches wherein the matrix uses zero masking when the first batch of individual datasets comprises less than one answer to each of the plurality of questions of the plurality of questionnaires. (Zhang [0096]: “The missing elements may be set to zero, 50% or some other predetermined value representing “no observation”. The value(s) of the corresponding features (i.e. same elements) of the feature space can then be read out from the decoded version {circumflex over (X)} of the feature vector, and taken as imputed values of the missing observations. In embodiments, the model 208′ may also be trained using some data points that have one or more missing values.”)
It would have been obvious before the effective filing date to combine self-attention based autoencoders of Purpura with the matrix masking of Zhang for efficient latent space use (Zhang, [0093]). Purpura and Zhang are analogous art because they both concern data imputation regarding autoencoders.
Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over Purpura and Kim and in further view of Le et al. (Supervised autoencoders: Improving generalization performance with unsupervised regularizers); hereinafter Le
Claim 10 is rejected over Purpura, Kim and Le with the incorporation of claim 1.
Regarding claim 10, Purpura does not appear to explicitly teach wherein the loss function comprises the sum of a choice prediction loss and a reconstruction loss, and wherein the choice prediction loss and the reconstruction loss are the mean squared error between a true target and a predicted output of the artificial intelligence model.
However, Le teaches wherein the loss function comprises the sum of a choice prediction loss and a reconstruction loss, and wherein the choice prediction loss and the reconstruction loss are the mean squared error between a true target and a predicted output of the artificial intelligence model. (Le [page 2, 2 Supervised autoencoders and representation learning]: “Let Lp be the supervised (primary) loss and Lr the loss for the reconstruction error. For example, in regression, both losses might be the squared error, resulting in the objective (1)”; Note: See Equation 1 to see that Lp is the choice prediction loss and Lr is the reconstruction loss.)
It would have been obvious before the effective filing date to combine self-attention based autoencoders of Purpura with the loss formula of Le to improve generalization (Le, page 2, 2 Supervised autoencoders and representation learning). Le and Purpura are analogous art because they both concern reconstructing data in autoencoders.
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Purpura and Kim and in further view of Somepalli et al. (SAINT: Improved Neural Networks for Tabular Data via Row Attention and Contrastive Pre-Training); hereinafter Somepalli
Claim 12 is rejected over Purpura, Kim and Somepalli with the incorporation of claim 1.
Regarding claim 12, Purpura does not appear to explicitly teach wherein the artificial intelligence model comprises a plurality of transformer blocks.
However, Somepalli teaches wherein the artificial intelligence model comprises a plurality of transformer blocks. (Somepalli [page 4, 3.1 Architecture]: “SAINT is composed of a stack of L identical stages. Each stage consists of one self-attention transformer block and one intersample attention transformer block. The self-attention transformer block is identical to the encoder from [41]. It has a multi-head self-attention layer (MSA) (with h heads), followed by two fully-connected feed-forward (FF) layers with a GELU non-linearity [16].”)
It would have been obvious before the effective filing date to combine self-attention based autoencoders of Purpura with the multiple transformer blocks of Somepalli to improve performance for tabular data (Somepalli, [page 1, 1 Introduction]). Somepalli and Purpura are analogous art because they both concern self-supervised reconstruction.
Claim 15 is rejected under 35 U.S.C. 103 as being unpatentable over Purpura and Kim and in further view of Kallianpur (US20220230024A1); hereinafter Kallianpur
Claim 15 is rejected over Purpura, Kim and Kallianpur with the incorporation of claim 1.
Regarding claim 15, Purpura teaches inputting the second batch of individual datasets into the artificial intelligence model; and (Purpura [page 8]: “The hyperparameters of our model are optimized separately on a random validation set—sampled each time from the training data—considering the following grid of points: number of items to mask during training n ∈ {1, 3, 5}, batch size b ∈ {16, 64, 128},”; and [page 8]: “we train the proposed model on the honest data from two of the three datasets with 10-fold cross validation and early stopping with patience 20—considering the MAE on the reconstruction in the validation set of each fold—and report the average performance the model achieved on the remaining dataset.”)
encoding the data of the second batch of individual datasets with the autoencoder. (Purpura [page 2]: “we propose to use an unsupervised deep learning approach based on an autoencoder. An autoencoder [3] is an unsupervised deep learning model that learns to reconstruct its input from a latent hidden representation. Given a set of answers to a questionnaire, we train the proposed model to reconstruct masked responses [4] in the input based on the remaining unmasked ones.”)
Purpura does not appear to explicitly teach determining if the artificial intelligence model exceeds a predetermined performance threshold; and
if the artificial intelligence model does not exceed the predetermined performance threshold:
generating a second batch of individual datasets;
However, Kallianpur teaches determining if the artificial intelligence model exceeds a predetermined performance threshold; and (Kallianpur [0079]: “At block 432, the process may determine whether the accuracy value exceeds an accuracy threshold. If yes, the process may proceed to block 440. If no, the process may proceed to block 434.”)
if the artificial intelligence model does not exceed the predetermined performance threshold:
generating a second batch of individual datasets; (Kallianpur [0024]: “If the accuracy value is below the given threshold, then a new model is created from the incoming data and used for further prediction.”; Note: See Figure 4 of Kallianpur and claim 9 to see “when the accuracy value does not exceed the accuracy threshold or when the similarity value does not exceed the similarity threshold, initiating a retraining process to generate a second ML model associated with the second data.”)
It would have been obvious before the effective filing date to combine self-attention based autoencoders of Purpura with the performance threshold of Kallianpur to improve the performance of the model (Kallianpur, [0059]). Kallianpur and Purpura are analogous art because they both concern optimizing machine learning models.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DAVID H TRAN whose telephone number is (703)756-1525. The examiner can normally be reached M-F 9:30 am - 5:30 pm.
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, Viker Lamardo can be reached at (571) 270-5871. 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.
/DAVID H TRAN/Examiner, Art Unit 2147
/VIKER A LAMARDO/Supervisory Patent Examiner, Art Unit 2147
Prosecution Insights