DETAILED ACTION
This action is written in response to the application filed 22 June 2023. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claim Rejections - 35 USC § 101
Claims 1, 3-4, 8, 10, 12-13, 17 and 19 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter. 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.
In determining whether the claims are subject matter eligible, the Examiner applies the 2019 USPTO Patent Eligibility Guidelines, as well as guidance from MPEP § 2106.
Step 1: Is the claim to a process, machine, manufacture, or composition of matter? Yes—claim 1 recites a method, which is a process.
Step 2A, prong one: Does the claim recite an abstract idea, law of nature or natural phenomenon? Yes—the claim recites one or more limitations which—under their broadest reasonable interpretation—covers performance of the limitation in the mind (see table below).
Claim limitation
Examiner analysis
1. A computer-implemented method for training an autoencoder to learn one or more chemical properties, comprising: …
optimizing the autoencoder using a loss function and the production rule sequence.
This is a mental process akin to a human evaluation.
Because the claim recites limitations which can practically be implemented as mental processes, the claim recites a mental process.
Step 2A, prong two: Does the claim recite additional elements that integrate the judicial exception into a practical application? No—the claim does not recite even generic computer hardware.
Step 2B: Does the claim recite additional elements that amount to significantly more than the judicial exception? No—the additional limitations are addressed below:
providing, as input, to an encoder of the autoencoder, a molecular graph representing a molecular structure;
This is insignificant pre-solution activity: inputting data to a model.
receiving, as output, from a decoder of the autoencoder, a production rule sequence for producing a molecule description of the molecular structure; and
This is insignificant post-solution / inter-solution activity: receiving results from a model.
The only limitation on the performance of the described method is that it must be “computer-implemented”. The claim thus recites computing components only at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using generic computer components. The statement that the method is performed by computer does not satisfy the test of “inventive concept.” See Alice Corp. Pty. Ltd. v. CLS Bank Int’l, 573 U.S. 208, 134 S. Ct. 2347, 2360 (2014).
For the reasons above, claim 1 is rejected as being directed to non-patentable subject matter under §101. This rejection applies equally to independent claims 10 and 19, which recite a system and a computer program product, respectively, as well as to dependent claims 3-4, 8, 12-13 and 17. The additional limitations of the dependent claims are addressed briefly below. Taken alone, the additional elements of the dependent claims above do not amount to significantly more than the above-identified judicial exception (the abstract idea). Looking at the limitations as an ordered combination adds nothing that is not already present when looking at the elements taken individually. There is no indication that the combination of elements improves the functioning of a computer or improves any other technology. Their collective functions merely provide conventional computer implementation.
Claim limitation
Examiner analysis
3. The computer-implemented method of claim 1, wherein the encoder comprises a first deep learning model, and the decoder comprises a second deep learning model.
This is merely additional information about one or more previously identified extra-solution models and/or mental processes.
4. The computer-implemented of claim 3, wherein the first deep learning model comprises a comprises a graph neural network (GNN), and the second deep learning model comprises a recurrent neural network (RNN).
This is merely additional information about one or more previously identified extra-solution models and/or mental processes.
8. The computer-implemented method of claim 1, wherein the loss function is a beta- variational autoencoder (beta-VAE) model.
This is merely additional information about one or more previously identified extra-solution models and/or mental processes.
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.
(b) the invention was patented or described in a printed publication in this or a foreign country or in public use or on sale in this country, more than one year prior to the date of application for patent in the United States.
Claims 1-3, 5-7, 10-12, 14-16, 19-20 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Kusner.
Kusner, Matt J., Brooks Paige, and José Miguel Hernández-Lobato. "Grammar variational autoencoder." In International conference on machine learning, pp. 1945-1954. PMLR, 2017.)
Regarding claims 1, 10 and 19, Kusner discloses a computer-implemented method (and a related system and computer program product) for training an autoencoder to learn one or more chemical properties, comprising:
providing, as input, to an encoder of the autoencoder, a molecular graph representing a molecular structure;
PNG
media_image1.png
288
1126
media_image1.png
Greyscale
P. 3, fig. 1, “The encoder of the GVAE.”
P. 3, sec. 3.1, “Encoding. Consider a subset of the SMILES grammar as shown in Figure 1, box 1 . These are the possible production rules that can be used for constructing a molecule. Imagine we are given as input the SMILES string for benzene: ‘c1ccccc1’. Figure 1, box 3 shows this molecule. To encode this molecule into a continuous latent representation we begin by using the SMILES grammar to parse this string into a parse tree (partially shown in box 2 ). This tree describes how ‘c1ccccc1’ is generated by the grammar. We decompose this tree into a sequence of production rules by performing a pre-order traversal on the branches of the parse tree from left-to-right, shown in box 4 . We convert these rules into 1-hot indicator vectors, where each dimension corresponds to a rule in the SMILES grammar, box 5 . These 1-hot vectors are concatenated into the rows of a matrix X of dimension T(X) ⇥K, where K is the number of production rules in the SMILES grammar, and T(X) is the number of production rules used to generate X.”
receiving, as output, from a decoder of the autoencoder, a production rule sequence for producing a molecule description of the molecular structure; and
PNG
media_image2.png
264
942
media_image2.png
Greyscale
P. 4, fig. 2, “The decoder of the GVAE.”
P. 3, sec. 3.1, “Decoding. We now describe how we map continuous vectors back to a sequence of production rules (and thus SMILES strings). Crucially we construct the decoder so that, at any time while we are decoding a sequence, the decoder will only be allowed to select a subset of production rules that are ‘valid’. This will cause the decoder to only produce valid parse sequences from the grammar.”
optimizing the autoencoder using a loss function and the production rule sequence.
P. 5, “Following Kingma & Welling (2014), we apply a noncentered parameterization on the encoding Gaussian distribution and optimize Eq. (4) using gradient descent, learning encoder and decoder neural network parameters.”
See also discussion of production rules throughout sec. 3.1.
Regarding independent claims 10 and 19, the additional computing components recited therein (ie “at least one processor”, “a non-transitory processor-readable memory”, and “a computer readable storage medium” are inherent throughout Kusner.
Regarding claims 2, 11 and 20, Kusner discloses the further limitation comprising:
receiving, as input, a training dataset comprising a first molecule description in line notation of a first molecular structure;
P. 3, sec. 3.2, ‘Training’. “During training, each input SMILES encoded as a sequence of 1-hot vectors X 2 {0, 1}Tmax⇥K, also defines a sequence of Tmax mask vectors.”
converting the first molecule description into the molecular graph, the molecular graph being a first molecular graph that represents the first molecular structure;
P. 3, fig. 1 (reproduced supra), step 2: “form parse tree”.
generating, based on the first molecular graph, a first production rule sequence of a context-free grammar (CFG) for the line notation;
P. 3, fig. 1 (reproduced supra), step 3: “extract rules”.
P. 4, first col., ‘CFG’ [context-free grammar].
encoding, via the encoder, the first molecular graph into a latent representation; and
P. 3, fig. 1 (reproduced supra), step 6: “map to latent spaces”.
decoding, via the decoder, the latent representation into a second production rule sequence of the CFG for the line notation, wherein the second production rule sequence is the production rule sequence received as output from the decoder;
P. 4, fig. 2 (reproduced supra), steps 2-4: “convert to logits”, “stack”, “mask out invalid rules”.
wherein the optimization of the autoencoder comprises:
performing a comparison of the first production rule sequence and the second production rule sequence in accordance with the loss function; and
P. 5, algorithm 2, “Training the Grammar VAE”.
training, based on the comparison, the encoder and the decoder.
Id.
Regarding claims 3 and 12, Kusner discloses the further limitation wherein the encoder comprises a first deep learning model, and the decoder comprises a second deep learning model.
PP. 3-4, figs. 1-2 (reproduced supra), illustrating an encoder and a decoder, respectively.
P. 3, sec. 3.1, “We use a deep convolutional neural network to map the collection of 1-hot vectors X to a continuous latent vector z.”.
P. 3, sec. 3.1, “We begin by passing the continuous vector z through a recurrent neural network”.
Regarding claims 5 and 14, Kusner discloses the further limitation wherein the line notation is simplified molecular- input line-entry system (SMILES).
P. 3, sec. 3.1, “SMILES”.
Regarding claims 6 and 15, Kusner discloses the further limitation comprising:
generating, based on the second production rule sequence, a second molecular graph representing a second molecular structure; and
P. 4, algorithm 2, describing the iterative training of the autoencoder using training examples until convergence is reached.
converting the second molecular graph into a second molecule description in the line notation of the second molecular structure.
Id. (See also mapping of claims 1 and 2 supra.)
Regarding claims 7 and 16, Kusner discloses the further limitation wherein the encoder and the decoder are trained to minimize a difference quantified by the loss function, such that the second molecule description is grammatically valid and describes a valid molecular structure.
P. 5, algorithm 2, “Update θ, φ … using gradient descent on the ELBO in Eq. (4)”. The Examiner notes that gradient descent necessarily relies upon a loss function.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103(a) which forms the basis for all obviousness rejections set forth in this Office action:
(a) A patent may not be obtained though the invention is not identically disclosed or described as set forth in section 102 of this title, if the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains. Patentability shall not be negatived by the manner in which the invention was made.
The following are the references relied upon in the rejections below:
Kusner (Kusner, Matt J., Brooks Paige, and José Miguel Hernández-Lobato. "Grammar variational autoencoder." In International conference on machine learning, pp. 1945-1954. PMLR, 2017.)
Li (Li, Zhen, Mingjian Jiang, Shuang Wang, and Shugang Zhang. "Deep learning methods for molecular representation and property prediction." Drug Discovery Today 27, no. 12 (2022): 103373.)
Polykovskiy (US 2021/0271980 A1)
Wang (Wang, Yuyang, Jianren Wang, Zhonglin Cao, and Amir Barati Farimani. "Molecular contrastive learning of representations via graph neural networks." Nature Machine Intelligence 4, no. 3 (2022): 279-287.)
Claims 4 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Kusner and Wang.
Regarding claims 4 and 13, Kusner discloses the further limitation wherein … the second deep learning model comprises a recurrent neural network (RNN).
P. 3, sec. 3.1, (cont.) “We begin by passing the continuous vector z through a recurrent neural network”.
Wang discloses the following further limitation which Kusner does not disclose:
wherein the first deep learning model comprises a comprises a graph neural network (GNN).
P. 2, “In this work, we propose MolCLR: Molecular Contrastive Learning of Representations via Graph Neural Networks shown in Figure 1 to address all the above challenges. …. Widely-used GNN models, Graph Convolutional Network (GCN)17 and Graph Isomorphism Network (GIN)18, are developed as GNN encoders in MolCLR to extract informative representation from molecule graphs.”
At the time of filing, it would have been obvious to a person of ordinary skill to apply a graph neural network (GNNs) for molecular structure encoding (as taught by Wang) in combination with the Kusner system because GNNs provide for a more complete and direct encoding of important topology information, as compared with string-based approaches like SMILES.
Claims 8 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Kusner and Polykovskiy.
Regarding claims 8 and 17, Polykovskiy discloses the further limitation which Kusner does not disclose wherein the loss function is a beta-variational autoencoder (beta-VAE) model.
[0086] “As an encoder and decoder, a 2-layer gated recurrent unit (GRU) network is used with a hidden size 128. The model is provided with a uniform prior and compare uniform and tricube proposals. For a baseline model, a β-VAE with Gaussian proposal and prior was trained.” (Emphasis added.)
At the time of filing, it would have been obvious to a person of ordinary skill to apply a β-VAE (as taught by Polykovskiy) to the Kusner system because it enables better interpretability than other VAE models. Both disclosures pertain to AE models for molecules.
Claims 9 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Kusner and Li.
Regarding claims 9 and 18, Li discloses the further limitation which Kusner does not disclose wherein the first production rule sequence is further based on molecular hypergraph grammar.
P. 6, “The generative method reconstructs the input through a encoder–decoder model. The molecular graph BERT63 combines local message passing and GNN into the BERT model for pre training. Koge et al.64 used a molecular hypergraph grammar variational autoencoder (VAE)65 to extract the embedding of molecules, which embedded the molecular structures and physical properties into the latent feature of VAE. The process of generative learning method based on molecular graphs is shown in Fig. 3b.” (Emphasis added.)
At the time of filing, it would have been obvious to a person of ordinary skill to apply a molecular hypergraph grammar (as taught by Li) to the Kusner system because it provides for complete encoding of important molecular topology information.
Additional Relevant Prior Art
The following references were identified by the Examiner as being relevant to the disclosed invention, but are not relied upon in any particular prior art rejection:
Kajino discloses the use of a molecular hypergraph grammar for use with a VAE. (Kajino, Hiroshi. "Molecular hypergraph grammar with its application to molecular optimization." In International Conference on Machine Learning, pp. 3183-3191. PMLR, 2019.)
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Vincent Gonzales whose telephone number is (571) 270-3837. The examiner can normally be reached on Monday-Friday 7 a.m. to 4 p.m. MT. If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Miranda Huang, can be reached at (571) 270-7092.
Information regarding the status of an application may be obtained from the USPTO 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.
/Vincent Gonzales/Primary Examiner, Art Unit 2124
Prosecution Insights