DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
Claims Status
Claims 1-15 are pending. Claims 1 and 8 are independent claims. Claims 1-15 are examined below.
Priority
As detailed on the 01/23/2023 filing receipt, this application does not claim priority. The effective filling date of this application is 12/28/2022.
Information Disclosure Statement
The Information Disclosure Statements filed 12/28/2022 is in compliance with the provisions of 37 CFR 1.97 and has therefore been considered. A signed copy of the IDS document is included with this Office Action.
Drawings
The drawings filed 12/28/2022 are accepted.
Claim Interpretation
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The claims in this application are given their broadest reasonable interpretation using the plain meaning of the claim language in light of the specification as it would be understood by one of ordinary skill in the art. The broadest reasonable interpretation of a claim element (also commonly referred to as a claim limitation) is limited by the description in the specification when 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is invoked.
As explained in MPEP § 2181, subsection I, claim limitations that meet the following three-prong test will be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph:
(A) the claim limitation uses the term “means” or “step” or a term used as a substitute for “means” that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function;
(B) the term “means” or “step” or the generic placeholder is modified by functional language, typically, but not always linked by the transition word “for” (e.g., “means for”) or another linking word or phrase, such as “configured to” or “so that”; and
(C) the term “means” or “step” or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function.
Use of the word “means” (or “step”) in a claim with functional language creates a rebuttable presumption that the claim limitation is to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites sufficient structure, material, or acts to entirely perform the recited function.
Absence of the word “means” (or “step”) in a claim creates a rebuttable presumption that the claim limitation is not to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is not interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites function without reciting sufficient structure, material or acts to entirely perform the recited function.
Claim limitations in this application that use the word “means” (or “step”) are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. Conversely, claim limitations in this application that do not use the word “means” (or “step”) are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action.
This application includes one or more claim limitations that do not use the word “means,” but are nonetheless being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the claim limitation(s) uses a generic placeholder that is coupled with functional language without reciting sufficient structure to perform the recited function and the generic placeholder is not preceded by a structural modifier.
Such claim limitation(s) is/are: “a system for translating an image of a structural formula of a chemical molecule into a textual identifier…” in claim 1.
Because this/these claim limitation(s) is/are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof.
If applicant does not intend to have this/these limitation(s) interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-7 and 10-11 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
See as discussed above (claim interpretation under 35 U.S.C. 112(f)), regarding claim 1, there is no disclosure in the originally filed specification of structure or material that is "A system for translating an image of a structural formula of a chemical molecule into a textual identifier" as presently claimed. Because there is no clear structure associated with the claimed terminology "a system" for performing the recited function, the claim is indefinite.
Claims 3-4 and 10-11 contains the trademark/trade name TensorFlow. Where a trademark or trade name is used in a claim as a limitation to identify or describe a particular material or product, the claim does not comply with the requirements of 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph. See Ex parte Simpson, 218 USPQ 1020 (Bd. App. 1982). The claim scope is uncertain since the trademark or trade name cannot be used properly to identify any particular material or product. A trademark or trade name is used to identify a source of goods, and not the goods themselves. Thus, a trademark or trade name does not identify or describe the goods associated with the trademark or trade name. In the present case, the trademark/trade name is used to identify/describe a machine learning library (spec. para. [0041]) and, accordingly, the identification/description is indefinite.
Dependent claims are rejected for depending on rejected claim.
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.
Analysis of claims in Step 1.
Step 1: Are the claims directed to a 101 process, machine, manufacture, or composition of matter (MPEP 2106.03)?
[Step 1: claims 1-15: NO]
Matter belonging to no 101 statutory category -- claims 1-15
Claims 1-15 are rejected under 35 USC 101 because the claimed inventions are directed to non-statutory subject matter.
Independent Claim 1 is to "a system," which is not, in all embodiments within a BRI, interpreted as belonging to any one particular category listed in 101. In a BRI, the claim reads on data and/or software comprising no structure other than data and/or software. The claim is not recited as a process, and the claim is not limited to any particular structure as a 101 machine or manufacture. The claim reads on transitory propagating signals which are not proper patentable subject matter because it does not fit within any of the four statutory categories of invention (In re Nuijten, Federal. Circuit, 2006).
In a BRI, none of the recited "database," is clearly a non-transitory element. In a BRI, each reads on information only in the form software and not clearly stored software.
In a BRI, none of the dependent claims 2-7 clearly requires a non-transitory element, and therefore none of the dependent claims clearly remedies this rejection.
Independent Claim 8 is to "computer readable storage medium," which is not, in all embodiments within a BRI, interpreted as belonging to any one particular category listed in 101. In a BRI, the claim reads on data and/or software comprising no structure other than data and/or software. The claim is not recited as a process, and the claim is not limited to any particular structure as a 101 machine or manufacture. The claim reads on transitory propagating signals which are not proper patentable subject matter because it does not fit within any of the four statutory categories of invention (In re Nuijten, Federal. Circuit, 2006).
In a BRI, none of the dependent claims 9-15 clearly requires a non-transitory element, and therefore none of the dependent claims clearly remedies this rejection.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claim(s) 1-3, 6, 8-10, 13 and 15 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Rajan ("Performance of chemical structure string representations for chemical image recognition using transformers." Digital Discovery 1.2 (2022): 84-90. First published on 15th January 2022; as cited on the attached 892 form).
Regarding independent claim 1, Rajan teaches a database configured to store unique tokens defined for each of known entities in chemical molecules with Table 2 (page 86). Table 2 lists databases for storing the chemical data and tokens.
Rajan teaches pre-process the image of the structural formula of the chemical molecule to generate a standardized image of the structural formula based on predefined parameters with “A production-quality bitmap image of each molecule was generated with the CDK Structure Diagram Generator (SDG) at a resolution of 300 300 pixels. Each molecule was rotated by a random angle ranging from 0 to 360 and depicted. The generated images were saved in 8 bit PNG format. Each image contains a single structure only.” (page 86, col. 1, para. 3).
Rajan teaches process the standardized image of the structural formula using an encoder-decoder architecture, wherein an encoder is implemented to generate embeddings for features in the standardized image of the structural formula and a decoder is implemented to utilize the generated embeddings along with an attention mechanism to associate each of the features in the standardized image of the structural formula to one of the unique tokens with “In this work, we use the same network as in DECIMER Image Transformer, a transformer-based network model similar to the “Base model” as explained in Google's publication, Attention Is All You Need.28 This network uses four encoder–decoder layers and eight attention heads. Attention has a dimension size of 512 and feed-forward networks have a dimension size of 2048. The columns and rows here correspond to the image features we extracted as vectors, which are 10 x 10 x 1536. A dropout rate of 10% is used to prevent overfitting. According to the publication “Attention Is All You Need” the network is trained using the Adam optimizer with a custom learning rate scheduler. The loss is calculated by using sparse categorical cross entropy between the real and predicted SELFIES. The network was coded with Python 3 using TensorFlow 2.3 (ref. 29) on the backend.” (page 86, col. 2, para. 3)
Rajan teaches recurrently process each of the features in the standardized image of the structural formula for predicting corresponding unique token based on the associated unique tokens therewith to generate multiple possible sequences complementary to the textual identifier, and dynamically calculate a correctness probability for each of the generated multiple possible sequences based on a confidence of each prediction of corresponding unique tokens involved therein with “The features from these images were extracted as vectors by using the pre-trained weights of the ‘noisy student’ trained EfficientNet-B3 (ref. 23) model. The extracted image features were then saved into NumPy arrays. These topics were discussed in detail in our previous publication.” (page 86, col. 1, para. 4) and “The extracted image features combined with the tokenized textual data were then converted into TFRecords. TFRecords are binary records that can be used to train a model faster using Cloud Tensor Processing Units (TPUs) on the Google Cloud Platform (GCP).” (page 86, col. 1, para. 5)
Rajan teaches select one of the multiple possible sequences with highest calculated correctness probability with “The purpose of this study was to examine different chemical string representations that are available for deep learning in chemistry by their performance on chemical image to string translation using transformer networks. Predictions were valid if the images could be translated into structures correctly.” (page 87, col. 1, para. 1)
Rajan teaches generate the textual identifier, as an output, for the image of the structural formula of the chemical molecule based on the selected one of the multiple possible sequences with " To further support OCSR development this work reports findings of a comparative case study for chemical image to chemical structure translation with SMILES, DeepSMILES and SELFIES. In addition, InChIs are included as an output which was proposed by a recent Kaggle competition.” (page 85, col. 1, para. 3)
Regarding claim 2, Rajan teaches implement a dataset of textual identifiers of known chemical molecules; compare the generated textual identifier to the textual identifiers of the known chemical molecules with “Once the models were fully converged, they were tested on an in-house server equipped with a GPU. To determine how many of the predictions were identical, the predictions were compared to the original strings. After the identical prediction calculations, all the predictions were converted to SMILES.” (page 86, col. 2, para. 6)
Rajan teaches determine if there is no match of the generated textual identifier to any one of the textual identifiers of the known chemical molecules with Tables 3 to 6 depicts the percentage of Identical predictions (string match) and Tanimoto 1.0 percentage (not identical) and “Identical predictions: this calculation identified how many predictions matched the original string representations. This was accomplished by using a one-to-one character string match. If a single character was wrong in the predicted string, it was considered as a wrong prediction.” (page 87, col. 1, para. 2, bullet 3). The recited “no match of the generated textual identifier” is interpreted to correspond to “Tanimoto 1.0 percentage (not identical)” as taught by Rajan.
Rajan teaches select one of the multiple possible sequences with next highest calculated correctness probability with “The purpose of this study was to examine different chemical string representations that are available for deep learning in chemistry by their performance on chemical image to string translation using transformer networks. Predictions were valid if the images could be translated into structures correctly.” (page 87, col. 1, para. 1) and “Valid SMILES: predicted SMILES and decoded SMILES which could be parsed to calculate the Tanimoto similarity calculations. The rest were classified as invalid SMILES.” (page 87, col. 1, para. 2, bullet 2). The recited “calculated correctness probability” is interpreted to correspond to “Tanimoto similarity calculations” as taught by Rajan.
Regarding claim 3, Rajan teaches wherein the processing arrangement implements a tensor workflow for the encoder-decoder architecture for processing the standardized image of the structural formula as a TensorFlow TFRecord with “The extracted image features combined with the tokenized textual data were then converted into TFRecords. TFRecords are binary records that can be used to train a model faster using Cloud Tensor Processing Units (TPUs) on the Google Cloud Platform (GCP).” (page 86, col. 1, para. 5) and “The network was coded with Python 3 using TensorFlow 2.3 (ref. 29) on the backend.” (page 86, col. 2, para. 3).
Regarding claim 6, Rajan teaches wherein the encoder implements one or more of: an EfficientNet encoder, an EfficientNetV2 encoder, a Vision Transformer (ViT) encoder, the decoder implements one or more of: a Recurrent Neural Network (RNN) with a Gated Recurrent Unit (GRU) decoder, a RNN with a Long Short-Term Memory (LSTM) decoder, a Transformer with self-attention decoder, the attention mechanism implements one or more of: Bahdanau Attention, Transformer self-attention with “The features from these images were extracted as vectors by using the pre-trained weights of the ‘noisy student’ trained EfficientNet-B3 model.” (page 86, col. 1, para. 4) and “In this work, we use the same network as in DECIMER Image Transformer, a transformer-based network model similar to the “Base model” as explained in Google's publication, Attention Is All You Need. This network uses four encoder–decoder layers and eight attention heads. Attention has a dimension size of 512 and feed-forward networks have a dimension size of 2048. The columns and rows here correspond to the image features we extracted as vectors, which are 10 x 10 x 1536. A dropout rate of 10% is used to prevent overfitting. According to the publication “Attention Is All You Need” the network is trained using the Adam optimizer with a custom learning rate scheduler. The loss is calculated by using sparse categorical cross entropy between the real and predicted SELFIES. The network was coded with Python 3 using TensorFlow 2.3 (ref. 29) on the backend.” (page 86, col. 2, para. 3)
Rajan teaches the predefined parameters comprise one or more of: a crop parameter for pre-processing the image of the structural formula of the chemical molecule, an aspect ratio parameter for pre-processing the image of the structural formula of the chemical molecule, a color inversion parameter for pre-processing the image of the structural formula of the chemical molecule with “A production-quality bitmap image of each molecule was generated with the CDK Structure Diagram Generator (SDG) at a resolution of 300 300 pixels. Each molecule was rotated by a random angle ranging from 0 to 360 and depicted. The generated images were saved in 8 bit PNG format. Each image contains a single structure only.” (page 86, col. 1, para. 3)
Rajan teaches the textual identifier, as the output, is one or more of: an International Chemical Identifier (InChI) textual identifier, Simplified molecular-input line-entry (SMILE) textual identifier, JSON files with each sublayer of the chemical notation as a separate field with “Valid DeepSMILES/SELFIES/InChI: the predicted Deep SMILES, SELFIES and InChIs that could decode back into SMILES strings.” (page 87, col. 1, para. 2, bullet 1)
Regarding independent claim 8, Rajan teaches a computer readable storage medium having computer executable instruction with “All investigations were performed using publicly available datasets and the code used to train and evaluate the models has been made available to the public.” (abstract).
Rajan teaches pre-process the image of the structural formula of the chemical molecule to generate a standardized image of the structural formula based on predefined parameters with “A production-quality bitmap image of each molecule was generated with the CDK Structure Diagram Generator (SDG) at a resolution of 300 300 pixels. Each molecule was rotated by a random angle ranging from 0 to 360 and depicted. The generated images were saved in 8 bit PNG format. Each image contains a single structure only.” (page 86, col. 1, para. 3).
Rajan teaches process the standardized image of the structural formula using an encoder-decoder architecture, wherein an encoder is implemented to generate embeddings for features in the standardized image of the structural formula and a decoder is implemented to utilize the generated embeddings along with an attention mechanism to associate each of the features in the standardized image of the structural formula to one of the unique tokens with “In this work, we use the same network as in DECIMER Image Transformer, a transformer-based network model similar to the “Base model” as explained in Google's publication, Attention Is All You Need.28 This network uses four encoder–decoder layers and eight attention heads. Attention has a dimension size of 512 and feed-forward networks have a dimension size of 2048. The columns and rows here correspond to the image features we extracted as vectors, which are 10 x 10 x 1536. A dropout rate of 10% is used to prevent overfitting. According to the publication “Attention Is All You Need” the network is trained using the Adam optimizer with a custom learning rate scheduler. The loss is calculated by using sparse categorical cross entropy between the real and predicted SELFIES. The network was coded with Python 3 using TensorFlow 2.3 (ref. 29) on the backend.” (page 86, col. 2, para. 3)
Rajan teaches recurrently process each of the features in the standardized image of the structural formula for predicting corresponding unique token based on the associated unique tokens therewith to generate multiple possible sequences complementary to the textual identifier, and dynamically calculate a correctness probability for each of the generated multiple possible sequences based on a confidence of each prediction of corresponding unique tokens involved therein with “The features from these images were extracted as vectors by using the pre-trained weights of the ‘noisy student’ trained EfficientNet-B3 (ref. 23) model. The extracted image features were then saved into NumPy arrays. These topics were discussed in detail in our previous publication.” (page 86, col. 1, para. 4) and “The extracted image features combined with the tokenized textual data were then converted into TFRecords. TFRecords are binary records that can be used to train a model faster using Cloud Tensor Processing Units (TPUs) on the Google Cloud Platform (GCP).” (page 86, col. 1, para. 5)
Rajan teaches select one of the multiple possible sequences with highest calculated correctness probability with “The purpose of this study was to examine different chemical string representations that are available for deep learning in chemistry by their performance on chemical image to string translation using transformer networks. Predictions were valid if the images could be translated into structures correctly.” (page 87, col. 1, para. 1)
Rajan teaches generate the textual identifier, as an output, for the image of the structural formula of the chemical molecule based on the selected one of the multiple possible sequences with " To further support OCSR development this work reports findings of a comparative case study for chemical image to chemical structure translation with SMILES, DeepSMILES and SELFIES. In addition, InChIs are included as an output which was proposed by a recent Kaggle competition.” (page 85, col. 1, para. 3)
Regarding claim 9, Rajan teaches implementing a dataset of textual identifiers of known chemical molecules; comparing the generated textual identifier to the textual identifiers of the known chemical molecules with “Once the models were fully converged, they were tested on an in-house server equipped with a GPU. To determine how many of the predictions were identical, the predictions were compared to the original strings. After the identical prediction calculations, all the predictions were converted to SMILES.” (page 86, col. 2, para. 6)
Rajan teaches determining if there is no match of the generated textual identifier to any one of the textual identifiers of the known chemical molecules with Tables 3 to 6 depicts the percentage of Identical predictions (string match) and Tanimoto 1.0 percentage (not identical) and “Identical predictions: this calculation identified how many predictions matched the original string representations. This was accomplished by using a one-to-one character string match. If a single character was wrong in the predicted string, it was considered as a wrong prediction.” (page 87, col. 1, para. 2, bullet 3). The recited “no match of the generated textual identifier” is interpreted to correspond to “Tanimoto 1.0 percentage (not identical)” as taught by Rajan.
Rajan teaches selecting one of the multiple possible sequences with next highest calculated correctness probability with “The purpose of this study was to examine different chemical string representations that are available for deep learning in chemistry by their performance on chemical image to string translation using transformer networks. Predictions were valid if the images could be translated into structures correctly.” (page 87, col. 1, para. 1) and “Valid SMILES: predicted SMILES and decoded SMILES which could be parsed to calculate the Tanimoto similarity calculations. The rest were classified as invalid SMILES.” (page 87, col. 1, para. 2, bullet 2). The recited “calculated correctness probability” is interpreted to correspond to “Tanimoto similarity calculations” as taught by Rajan.
Regarding claim 10, Rajan teaches wherein the processing the standardized image of the structural formula as a TensorFlow TFRecord with “The extracted image features combined with the tokenized textual data were then converted into TFRecords. TFRecords are binary records that can be used to train a model faster using Cloud Tensor Processing Units (TPUs) on the Google Cloud Platform (GCP).” (page 86, col. 1, para. 5) and “The network was coded with Python 3 using TensorFlow 2.3 (ref. 29) on the backend.” (page 86, col. 2, para. 3).
Regarding claim 13, Rajan teaches wherein the encoder implements one or more of: an EfficientNet encoder, an EfficientNetV2 encoder, a Vision Transformer (ViT) encoder, the decoder implements one or more of: a Recurrent Neural Network (RNN) with a Gated Recurrent Unit (GRU) decoder, a RNN with a Long Short-Term Memory (LSTM) decoder, a Transformer with self-attention decoder, the attention mechanism implements one or more of: Bahdanau Attention, Transformer self-attention with “The features from these images were extracted as vectors by using the pre-trained weights of the ‘noisy student’ trained EfficientNet-B3 model.” (page 86, col. 1, para. 4) and “In this work, we use the same network as in DECIMER Image Transformer, a transformer-based network model similar to the “Base model” as explained in Google's publication, Attention Is All You Need. This network uses four encoder–decoder layers and eight attention heads. Attention has a dimension size of 512 and feed-forward networks have a dimension size of 2048. The columns and rows here correspond to the image features we extracted as vectors, which are 10 x 10 x 1536. A dropout rate of 10% is used to prevent overfitting. According to the publication “Attention Is All You Need” the network is trained using the Adam optimizer with a custom learning rate scheduler. The loss is calculated by using sparse categorical cross entropy between the real and predicted SELFIES. The network was coded with Python 3 using TensorFlow 2.3 (ref. 29) on the backend.” (page 86, col. 2, para. 3)
Rajan teaches the predefined parameters comprise one or more of: a crop parameter for pre-processing the image of the structural formula of the chemical molecule, an aspect ratio parameter for pre-processing the image of the structural formula of the chemical molecule, a color inversion parameter for pre-processing the image of the structural formula of the chemical molecule with “A production-quality bitmap image of each molecule was generated with the CDK Structure Diagram Generator (SDG) at a resolution of 300 300 pixels. Each molecule was rotated by a random angle ranging from 0 to 360 and depicted. The generated images were saved in 8 bit PNG format. Each image contains a single structure only.” (page 86, col. 1, para. 3)
Rajan teaches the textual identifier, as the output, is one or more of: an International Chemical Identifier (InChI) textual identifier, Simplified molecular-input line-entry (SMILE) textual identifier, JSON files with each sublayer of the chemical notation as a separate field with “Valid DeepSMILES/SELFIES/InChI: the predicted Deep SMILES, SELFIES and InChIs that could decode back into SMILES strings.” (page 87, col. 1, para. 2, bullet 1)
Regarding claim 15, Rajan teaches a computer program comprising computer executable program code, when executed the computer executable program code controls a computer system to perform the method according to claim 8 with “All investigations were performed using publicly available datasets and the code used to train and evaluate the models has been made available to the public.” (abstract).
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 4 and 11 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rajan ("Performance of chemical structure string representations for chemical image recognition using transformers." Digital Discovery 1.2 (2022): 84-90. First published on 15th January 2022; as cited on the attached 892 form) as applied to claims 1-3, 6, 8-10, 13 and 15 as indicated in the 35 U.S.C. 102(a)(1) section above; in view of Rand (US 2022/0044149 A1, published Feb. 10, 2022; as cited on the attached 892 form).
Rajan is applied to claims 1-3, 6, 8-10, 13 and 15 as indicated in the 35 U.S.C. 102(a)(1) section above.
Rajan does not explicitly teach wherein the encoderis configured to implement a mixed precision accuracy scheme to the embeddings for the features in the TensorFlow TFRecord in claims 4 and 11. However, these limitations are taught by Rand.
Regarding claim 4, Rand teaches wherein the encoder, in the processing arrangement, is configured to implement a mixed precision accuracy scheme to the embeddings for the features in the TensorFlow TFRecord with “It is also critical to have the tools for in-depth analysis of the training pipeline. These tools should be built around basic tools for measuring resource utilization (e.g. nvidia-smi or the Sagemaker instance metrics) and tf profiler (https://www.tensorflow.org/guide/profiler), for profiling your model. There are also many techniques for improving performance (e.g. mixed precision (https://www.tensorflow.org/guide/keras/mixed_precision)). The techniques implemented should be dictated by the profiling data.” [0181].
Regarding claim 11, Rand teaches wherein the encoder is configured to implement a mixed precision accuracy scheme to the embeddings for the features in the TensorFlow TFRecord with “It is also critical to have the tools for in-depth analysis of the training pipeline. These tools should be built around basic tools for measuring resource utilization (e.g. nvidia-smi or the Sagemaker instance metrics) and tf profiler (https://www.tensorflow.org/guide/profiler), for profiling your model. There are also many techniques for improving performance (e.g. mixed precision (https://www.tensorflow.org/guide/keras/mixed_precision)). The techniques implemented should be dictated by the profiling data.” [0181].
It would have been prima facia obvious to combine the teachings of Rajan and Rand to arrive at the claimed invention. A person of ordinary skill in the art would have been motivated to modify the method of Rajan to include a mixed precision accuracy scheme as taught by Rand to improve the performance of the model. Furthermore, there would have been a reasonable expectation of success, since Rajan and Rand teach methods that pertain to the generation of machine learning models with TensorFlow.
Claim(s) 5 and 12 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rajan ("Performance of chemical structure string representations for chemical image recognition using transformers." Digital Discovery 1.2 (2022): 84-90. First published on 15th January 2022; as cited on the attached 892 form) as applied to claims 1-3, 6, 8-10, 13 and 15 as indicated in the 35 U.S.C. 102(a)(1) section above in view of Musazade (Review of techniques and models used in optical chemical structure recognition in images and scanned documents. J Cheminform 14, 61 (2022). Published 09 September 2022; as cited on the attached 892 form).
Rajan is applied to claims 1-3, 6, 8-10, 13 and 15 as indicated in the 35 U.S.C. 102(a)(1) section above.
Rajan does not explicitly teach wherein the processing arrangement is configured to implement a beam search technique for the recurrent processing of each of the features in the standardized image of the structural formula for predicting corresponding unique token based on the associated unique tokens therewith to generate multiple possible sequences complementary to the textual identifier in claims 5 and 12. However, these limitations are taught by Musazade.
Regarding claim 5, Musazade teaches wherein the processing arrangement is configured to implement a beam search technique for the recurrent processing of each of the features in the standardized image of the structural formula for predicting corresponding unique token based on the associated unique tokens therewith to generate multiple possible sequences complementary to the textual identifier with “Beam search. Instead of sampling once at each step, multiple word sequences are selected and kept as candidate sequences at every time step. The number of candidates is predefined by the k parameter which is the beam. The final outputted sequence is the one with the highest total log probability over all generated characters. This is better than the greedy approach because it prevents the model from being stuck due to a bad decision at some stage of sequence prediction.” (Page 11, col. 1, para. 3 to Page 11, col. 2, para. 1).
Regarding claim 12, Musazade teaches implementing a beam search technique for the recurrent processing of each of the features in the standardized image of the structural formula for predicting corresponding unique token based on the associated unique tokens therewith to generate multiple possible sequences complementary to the textual identifier with “Beam search. Instead of sampling once at each step, multiple word sequences are selected and kept as candidate sequences at every time step. The number of candidates is predefined by the k parameter which is the beam. The final outputted sequence is the one with the highest total log probability over all generated characters. This is better than the greedy approach because it prevents the model from being stuck due to a bad decision at some stage of sequence prediction.” (Page 11, col. 1, para. 3 to Page 11, col. 2, para. 1)
It would have been prima facia obvious to combine the teachings of Rajan and Musazade to arrive at the claimed invention. A person of ordinary skill in the art would have been motivated to modify the method of Rajan to include beam search as taught by Musazade to prevents the model from being stuck due to a bad decision at some stage of sequence prediction. Furthermore, there would have been a reasonable expectation of success, since Rajan and Musazade teach methods that pertain to the generation of machine learning models.
Claim(s) 7 and 14 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rajan ("Performance of chemical structure string representations for chemical image recognition using transformers." Digital Discovery 1.2 (2022): 84-90. First published on 15th January 2022; as cited on the attached 892 form) as applied to claims 1-3, 6, 8-10, 13 and 15 as indicated in the 35 U.S.C. 102(a)(1) section above in view of Klemm ("Barista-a graphical tool for designing and training deep neural networks." arXiv preprint arXiv:1802.04626 (2018).; as cited on the attached 892 form).
Rajan is applied to claims 1-3, 6, 8-10, 13 and 15 as indicated in the 35 U.S.C. 102(a)(1) section above.
Rajan does not explicitly teach provided via a user-interface, to allow a user to select a combination of one of the encoder, one of the decoder, one of the attention mechanism, one of the predefined parameters, and one of the textual identifier as the output in claim 7. However, this limitation is taught by Klemm.
Regarding claim 7, Klemm teaches a user-interface, to allow a user to select a combination of one of the encoder, one of the decoder, one of the attention mechanism, one of the predefined parameters, and one of the textual identifier as the output with “The network editor allows direct manipulation of the network, like adding or deleting layers and editing connections, without the need to type a single line of prototext definition.” (page 4, para. 1) and “Further settings are controlled using different docks, which can be activated and positioned freely by the user. Parameters of a specific layer can be set using the layer properties dock. Here, the user is provided with a list of all available parameter groups and parameters for the selected layer. Depending on the parameter type a list of valid settings is shown, further reducing the need to refer to the Caffe documentation for available options. Hyper parameters to control training of the NN can be set using a similar interface in the solver properties dock. A list of all layer types of the selected Caffe branch is available in the layers dock. New layers can be added to the network editor using Drag&Drop. Editing larger networks is simplified by a text search provided in the network layers dock.” (page 4, para. 2)
Regarding claim 14, Klemm teaches a user-interface, to allow a user to select a combination of one of the encoder, one of the decoder, one of the attention mechanism, one of the predefined parameters, and one of the textual identifier as the output with “The network editor allows direct manipulation of the network, like adding or deleting layers and editing connections, without the need to type a single line of prototext definition.” (page 4, para. 1) and “Further settings are controlled using different docks, which can be activated and positioned freely by the user. Parameters of a specific layer can be set using the layer properties dock. Here, the user is provided with a list of all available parameter groups and parameters for the selected layer. Depending on the parameter type a list of valid settings is shown, further reducing the need to refer to the Caffe documentation for available options. Hyper parameters to control training of the NN can be set using a similar interface in the solver properties dock. A list of all layer types of the selected Caffe branch is available in the layers dock. New layers can be added to the network editor using Drag&Drop. Editing larger networks is simplified by a text search provided in the network layers dock.” (page 4, para. 2)
It would have been prima facia obvious to combine the teachings of Rajan and Klemm to arrive at the claimed invention. A person of ordinary skill in the art would have been motivated to modify the method of Rajan to include user-interface as taught by Klemm to allow users to modify the model to better suit their requirements. Furthermore, there would have been a reasonable expectation of success, since Rajan and Klemm teach methods that pertain to the generation of machine learning models.
Conclusion
No claims are allowed.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KETTIP KRIANGCHAIVECH whose telephone number is (571)272-1735. The examiner can normally be reached 8:30am-5:00pm EDT.
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, Larry D. Riggs can be reached at (571) 270-3062. 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.
/K.K./Examiner, Art Unit 1686
/LARRY D RIGGS II/Supervisory Patent Examiner, Art Unit 1686
Prosecution Insights