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 .
DETAILED ACTION
Note: The claims are not directed towards patent ineligible subject matter under 35 U.S.C. 101
Step 1: IS THE CLAIM DIRECTED TO A PROCESS, MACHINE, MANUFACTURE OR COMPOSITION OF MATTER?
Yes
Step 2A.1: IS THE CLAIM DIRECTED TO A LAW OF NATURE, A NATURAL PHENOMENON (PRODUCT OF NATURE) OR AN ABSTRACT IDEA?
No
Step 2A.2: DOES THE CLAIM RECITE ADDITIONAL ELEMENTS THAT INTEGRATE THE JUDICIAL EXCEPTION INTO A PRACTICAL APPLICATION?
Yes, if the claims are alternatively construed to be abstract in step 2A1. The claims seek to improve a LLM performance and reduction of memory supported by the specification, and reflected by the claims e.g. in spec: 0044 0047 In other words, the claims enable the invention to use decoding and machine learning to utilize projections which provide improved speculative decoding used in machine learning models (e.g., transformer-based models such as LLMs), resulting in reduced memory consumption in speculative decoding.
Supported by the following:
In Finjan Inc. v. Blue Coat Systems, Inc., 879 F.3d 1299, 125 USPQ2d 1282 (Fed. Cir. 2018), the claimed invention was a method of virus scanning that scans an application program, generates a security profile identifying any potentially suspicious code in the program, and links the security profile to the application program. 879 F.3d at 1303-04, 125 USPQ2d at 1285-86. The Federal Circuit noted that the recited virus screening was an abstract idea, and that merely performing virus screening on a computer does not render the claim eligible. 879 F.3d at 1304, 125 USPQ2d at 1286. The court then continued with its analysis under part one of the Alice/Mayo test by reviewing the patent’s specification, which described the claimed security profile as identifying both hostile and potentially hostile operations. The court noted that the security profile thus enables the invention to protect the user against both previously unknown viruses and “obfuscated code,” as compared to traditional virus scanning, which only recognized the presence of previously-identified viruses. The security profile also enables more flexible virus filtering and greater user customization. 879 F.3d at 1304, 125 USPQ2d at 1286. The court identified these benefits as improving computer functionality, and verified that the claims recite additional elements (e.g., specific steps of using the security profile in a particular way) that reflect this improvement. Accordingly, the court held the claims eligible as not being directed to the recited abstract idea. 879 F.3d at 1304-05, 125 USPQ2d at 1286-87. This analysis is equivalent to the Office’s analysis of determining that the additional elements integrate the judicial exception into a practical application at Step 2A Prong Two, and thus that the claims were not directed to the judicial exception (Step 2A: NO).
Examples of claims that improve technology and are not directed to a judicial exception include: Enfish, LLC v. Microsoft Corp., 822 F.3d 1327, 1339, 118 USPQ2d 1684, 1691-92 (Fed. Cir. 2016) (claims to a self-referential table for a computer database were directed to an improvement in computer capabilities and not directed to an abstract idea); McRO, Inc. v. Bandai Namco Games Am. Inc., 837 F.3d 1299, 1315, 120 USPQ2d 1091, 1102-03 (Fed. Cir. 2016) (claims to automatic lip synchronization and facial expression animation were directed to an improvement in computer-related technology and not directed to an abstract idea); Visual Memory LLC v. NVIDIA Corp., 867 F.3d 1253,1259-60, 123 USPQ2d 1712, 1717 (Fed. Cir. 2017) (claims to an enhanced computer memory system were directed to an improvement in computer capabilities and not an abstract idea); Finjan Inc. v. Blue Coat Systems, Inc., 879 F.3d 1299, 125 USPQ2d 1282 (Fed. Cir. 2018) (claims to virus scanning were found to be an improvement in computer technology and not directed to an abstract idea); SRI Int’l, Inc. v. Cisco Systems, Inc., 930 F.3d 1295, 1303 (Fed. Cir. 2019) (claims to detecting suspicious activity by using network monitors and analyzing network packets were found to be an improvement in computer network technology and not directed to an abstract idea). Additional examples are provided in MPEP § 2106.05(a).
Regarding the December 5th 2025 Memo in light of September 26, 2025 Appeals Review Panel Decision in Ex parte Desjardins, Appeal 2024-000567 for Application 16/319,040, in deciding if a recited abstract idea does or does not direct the entire claim to an abstract idea, when a claim is considered as a whole:
Paragraph 21 of the Specification, which the Appellant cites, identifies improvements in training the machine learning model itself. Of course, such an assertion in the Specification alone is insufficient to support a patent eligibility determination, absent a subsequent determination that the claim itself reflects the disclosed improvement. See MPEP § 2106.05(a) (citing Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1316 (Fed. Cir. 2016)). Here, however, we are persuaded that the claims reflect such an improvement. For example, one improvement identified in the 8 Appeal2024-000567 Application 16/319,040 Specification is to "effectively learn new tasks in succession whilst protecting knowledge about previous tasks." Spec. ,r 21. The Specification also recites that the claimed improvement allows artificial intelligence (AI) systems to "us[e] less of their storage capacity" and enables "reduced system complexity." Id. When evaluating the claim as a whole, we discern at least the following limitation of independent claim 1 that reflects the improvement: "adjust the first values of the plurality of parameters to optimize performance of the machine learning model on the second machine learning task while protecting performance of the machine learning model on the first machine learning task." We are persuaded that constitutes an improvement to how the machine learning model itself operates, and not, for example, the identified mathematical calculation. Under a charitable view, the overbroad reasoning of the original panel below is perhaps understandable given the confusing nature of existing § 101 jurisprudence, but troubling, because this case highlights what is at stake. Categorically excluding AI innovations from patent protection in the United States jeopardizes America's leadership in this critical emerging technology. Yet, under the panel's reasoning, many AI innovations are potentially unpatentable-even if they are adequately described and nonobvious-because the panel essentially equated any machine learning with an unpatentable "algorithm" and the remaining additional elements as "generic computer components," without adequate explanation. Dec. 24. Examiners and panels should not evaluate claims at such a high level of generality.
Specifically, Ex Parte Desjardins explained the following:
Enfish ranks among the Federal Circuit's leading cases on the eligibility of technological improvements. In particular, Enfish recognized that “[m]uch of the advancement made in computer technology consists of improvements to software that, by their very nature, may not be defined by particular physical features but rather by logical structures and processes.” 822 F.3d at 1339. Moreover, because “[s]oftware can make non-abstract improvements to computer technology, just as hardware improvements can,” the Federal Circuit held that the eligibility determinations should turn on whether “the claims are directed to an improvement to computer functionality versus being directed to an abstract idea.” Id. at 1336. (Desjardins, page 8).
Further in Ex Parte Desjardins, Appeal No. 2024-000567 (PTAB September 26, 2025, Appeals Review Panel Decision) (precedential), the claimed invention was a method of training a machine learning model on a series of tasks. The Appeals Review Panel (ARP) overall credited benefits including reduced storage, reduced system complexity and streamlining, and preservation of performance attributes associated with earlier tasks during subsequent computational tasks as technological improvements that were disclosed in the patent application specification. Specifically, the ARP upheld the Step 2A Prong One finding that the claims recited an abstract idea (i.e., mathematical concept). In Step 2A Prong Two, the ARP then determined that the specification identified improvements as to how the machine learning model itself operates, including training a machine learning model to learn new tasks while protecting knowledge about previous tasks to overcome the problem of “catastrophic forgetting” encountered in continual learning systems. Importantly, the ARP evaluated the claims as a whole in discerning at least the limitation “adjust the first values of the plurality of parameters to optimize performance of the machine learning model on the second machine learning task while protecting performance of the machine learning model on the first machine learning task” reflected the improvement disclosed in the specification. Accordingly, the claims as a whole integrated what would otherwise be a judicial exception instead into a practical application at Step 2A Prong Two, and therefore the claims were
The claim itself does not need to explicitly recite the improvement described in the specification (e.g., “thereby increasing the bandwidth of the channel”). See, e.g., Ex Parte Desjardins, Appeal No. 2024-000567 (PTAB September 26, 2025, Appeals Review Panel Decision) (precedential), in which the specification identified the improvement to machine learning technology by explaining how the machine learning model is trained to learn new tasks while protecting knowledge about previous tasks to overcome the problem of “catastrophic forgetting,” and that the claims reflected the improvement identified in the specification. Indeed, enumerated improvements identified in the Desjardins specification included disclosures of the effective learning of new tasks in succession in connection with specifically protecting knowledge concerning previously accomplished tasks; allowing the system to reduce use of storage capacity; and the enablement of reduced complexity in the system. Such improvements were tantamount to how the machine learning model itself would function in operation and therefore not subsumed in the identified mathematical calculation.
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 –
(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.
Claims 1, 2, 9, 10, 17, and 18 rejected under 35 U.S.C. 102(a)(1) as being anticipated by US 12087306 B1 Le; Duc Hoang et al. (hereinafter Le).
Re claim 1, Le teaches
1. An apparatus for using a machine learning model to generate tokens, the apparatus comprising: (fig. 6)
at least one memory; and at least one processor coupled to the at least one memory and configured to: (fig. 6)
process, using a linear layer of a machine learning model, an embedding generated from a first output token and input features to generate first features, wherein the first output token is generated by a previous iteration of a token predictor and wherein the input features are generated by a previous iteration of a decoding layer; (in fig. 3 with col 6 line 58 to col 8 line 29, a reduction output from element 310 takes place with bias embedding using previous tokens fed into a joiner/linear hidden layer blocks from a decoder, each layer is trained e.g. cold fusion for RNNT and NNLM)
process, using the decoding layer, the first features to generate second features having first dimensions; (the decoder in fig. 3 with col 6 line 58 to col 8 line 29, a reduction output from element 310 takes place with bias embedding using previous tokens fed into a joiner/linear hidden layer blocks from a decoder, each layer is trained e.g. cold fusion for RNNT and NNLM)
process, using a down-projection layer, the second features to generate third features having second dimensions smaller than the first dimensions; and (reduction of dimension in fig. 3 with col 6 line 58 to col 8 line 29, a reduction output from element 310 takes place with bias embedding using previous tokens fed into a joiner/linear hidden layer blocks from a decoder, each layer is trained e.g. cold fusion for RNNT and NNLM)
generate, using the token predictor and the third features, a second output token. (new token predicted from past features in fig. 3 with col 6 line 58 to col 8 line 29, a reduction output from element 310 takes place with bias embedding using previous tokens fed into a joiner/linear hidden layer blocks from a decoder, each layer is trained e.g. cold fusion for RNNT and NNLM)
Re claim 9, this claim has been rejected for teaching a broader, or narrower claim based on general inclusion of hardware alone (e.g. processor, memory, instructions), representation of claim 1 omitting/including hardware for instance, otherwise amounting to a virtually identical scope
For instance, see fig. 6 which contains the hardware/software.
Re claim 17, this claim has been rejected for teaching a broader, or narrower claim based on general inclusion of hardware alone (e.g. processor, memory, instructions), representation of claim 1 omitting/including hardware for instance, otherwise amounting to a virtually identical scope
For instance, see fig. 6 which contains the hardware/software.
Re claims 2, 10, and 18, Le teaches
2. The apparatus of claim 1, wherein the linear layer comprises first parameters, the decoding layer comprises second parameters, and the down-projection layer comprises third parameters. (linear is the joiner, in fig. 3 with col 6 line 58 to col 8 line 29, a reduction output from element 310 takes place with bias embedding using previous tokens fed into a joiner/linear hidden layer blocks from a decoder, each layer is trained e.g. cold fusion for RNNT and NNLM)
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 8 and 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over US 12087306 B1 Le; Duc Hoang et al. (hereinafter Le) in view of US 20250173392 A1 CHOI; Chiho et al. (hereinafter CHOI).
Re claims 8 and 16, while Le teaches RNNT and NNLN with extremely large amounts of data, in lieu of official notice, it fails to teach:
8. The apparatus of claim1, wherein the machine learning model is a large language model (LLM). (CHOI 0080 LLM)
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the system of Le to incorporate the above claim limitations as taught by CHOI to allow for simple substitution of one known element for another to obtain predictable results such as LLM in place of NNLM or RNNT which both analogously use large amounts of data for training, thereby updating the NN concepts with a more modem model e.g. LLM, for improved context retention, adaptability, and reasoning capabilities.
Allowable Subject Matter
Claims 3-7, 11-15, and 19-20:
Are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims. After searching through patent and non-patent literature, there was no evidence that there exists a limitation in direct relation or an obvious variant to such limitations as a whole as precisely limited. When searching for a secondary prior art for the limitation as recited in the above claims, the most relevant topics pertained to material from the same Inventor and Assignee but did not teach or suggest the aforementioned complex limitations as a whole as precisely limited.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US 12277409 B1 Leeman-Munk; Samuel Paul et al.
High low dimension projection with tokenization
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MICHAEL C COLUCCI whose telephone number is (571)270-1847. The examiner can normally be reached on M-F 9 AM - 5 PM.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Andrew Flanders can be reached at (571)272-7516. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/MICHAEL COLUCCI/Primary Examiner, Art Unit 2655 (571)-270-1847
Examiner FAX: (571)-270-2847
Michael.Colucci@uspto.gov
Prosecution Insights