DETAILED ACTION
This action is responsive to the Remarks filed on 10/14/2025. Claims 1, 3-9, 11-17, 19-20 are pending in the case. Claims 1, 9, and 17 are independent claims.
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 .
Response to Arguments
Applicant's arguments filed 10/14/2025 with respect to the 35 U.S.C 101 rejection has been considered but is not persuasive.
With respect to 101 Step 2A prong 1
Applicant appears not to dispute that the claims recite at least one mental process, instead simply states that the claims recite limitations that cannot be performed in the mind.
As previously noted, this step as described in the MPEP asks the Examiner to identify at least one limitation which recites an abstract idea (see MPEP 2106.04(a)). The consideration of other limitations which do not fall within the enumerated abstract idea categories is not part of this prong. As such the statement by Applicant “the claims recite limitations that cannot practically be performed in the human mind” does not address the contention that the claim in fact recites certain specific limitations identified in the rejection as “practically performed in the human mind”
With respect to 101 Step 2A prong II
Applicant argues the claims provide an improvement (i.e reducing power/memory consumption of a computer. Applicant cites paragraph 0027 of spec which alludes to the method allowing for reduction in the number of models required for language pairing. Applicant emphasizes several elements of the independent claim which they believe enable the improvement. In particular, certain emphasized features indicate how the decoder neural network is shared across languages (“…generating a tokenized input…using a shared vocabulary…prepending an identifier token…an identifier that specifies the candidate language…wherein the machine learning model has been trained”). Further such features do not merely recite “applying the model” but technological features that enable the improvement.
Finally, Applicant nots that the improvement is not the result of the abstract idea alone but of specific “logical structures and processes” which provide a particular solution.
Examiner disagrees. Critically, each of these features only serve to nominally describe the type of data evaluated in the abstract idea (i.e tokenizing using a shared vocabulary can be performed in the mind, prepending particular tokens to existing tokens is a method for arranging and organizing abstract information in the mind). Merely applying an abstract idea to an input of machine learning technology such as a “trained machine learning model” does not reflect an additional element which is indicative of an improvement. The improvement described in the specification is directly a result of the preprocessing and data manipulation which amounts to the identified abstract idea. Applicant’s argument that the claim recites specific features which recite “how the decoder neural network is shared” appear to place emphasis on the identified abstract ideas alone (see pg 10 of 12 of the remarks.)
Examiner notes it is therefore reasonable to conclude the “how” is a result of the abstract ideas alone, any improvement should be reflected in the claim via the additional elements and should not be the result of an abstract idea alone. As noted in the rejection, the claims describe additional elements which are described in the MPEP as elements which do not integrate into a practical application or significantly more and thus cannot be considered an improvement (see 2106.05(f-h) and 2106.05(d)). The use previously trained model on certain data whose performance is better than other models does not provide an improvement in computer technology for the same reasons that using a super-computer rather than an office computer for predicting the weather is not an improvement in computer technology. Similarly, adding details which generally link to a particular field such as the type of weather data used or the type of algorithms used would not proport to improve the functioning of the super-computer for the same reason that the details of the training data or the task (i.e. identifier tokens and shared vocabulary) do not improve the functioning of the neural network. At most the machine learning details recited claim the idea of a solution without actively claiming specific training steps which reflect the improvement.
For the above reasons, Applicant’s concluding statement that the improvement is the result of specific “logical structures” is not convincing.
For the above reasons the rejection is maintained.
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, 3-9, 11-17, 19-20 are rejected under 35 U.S.C. 101 because the claims are directed to an abstract idea without significantly more.
Regarding Claim 1, 9 and 17
Under step 1, claim 1 is directed to a computer-implemented method, which is directed to a process, one of the statutory categories. Under step 1, claim 9 is directed to A system, which is directed to a machine, one of the statutory categories. Under step 1, claim 17 is directed to One or more non-transitory computer-readable storage media storing instructions which is directed to a product of manufacture, one of the statutory categories.
Under Step 2A Prong 1, the claim recites the following limitations which are considered mental evaluations “selecting from a plurality of candidate task identifiers that each identifies a respective language processing task, a task identifier that identifies the target task specified in the input…generating a tokenized input by tokenizing the language input using a shared vocabulary that includes tokens representing sub-word units that are shared between a plurality of different candidate languages…Processing the tokenized input and the selected task identifier …to generate a language output… to encode the model input independently of the target task;… to generate outputs from a shared vocabulary and wherein processing the tokenized input and the selected task identifier comprises prepending the selected task identifier identifying the target task specified in the input to an output of the decoder neural network.” Examiner notes that the generation of a language output amounts to selecting a translation of a language. Encoding input may involve representing characters as corresponding numerical values, such as ASCII encoding. Output from a shared vocabular may be a selection of characters from the roman alphabet shared among a plurality of languages. Prepending the task identifier amounts to selecting which language an output belongs to. Each of these steps can be performed in the human mind as they are all evaluations about abstract data.
Therefore, the claim recites an abstract idea.
Step 2A Prong Two Analysis: The judicial exception in not integrated into a practical application. In particular, the claims recite the additional element(s) the limitations “an encoder neural network… and a decoder neural network that is shared between the plurality of different candidate languages… wherein the machine learning model has been trained on a set of training examples comprising for each candidate language in a plurality of different candidate languages, a respective set of training examples that each includes (i) a training language input in a respective first language, (ii) a training language output in the candidate language, and (iii) an identifier that specifies the candidate language, … using a machine learning model… one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising:… instructions that when executed by one or more computers cause the one or more computers to perform operations comprising:” amounts to mere instructions to apply a computer technology to an abstract idea at a high degree of generality, see MPEP 2106.05(f). None of the additional elements describe how the model is applied, but only that certain named models are used to perform the above recited abstract ideas. Alternatively, these additional elements “wherein the machine learning model has been trained on a set of training examples comprising for each candidate language in a plurality of different candidate languages, a respective set of training examples that each includes (i) a training language input in a respective first language, (ii) a training language output in the candidate language, and (iii) an identifier that specifies the candidate language” may be considered limitations which amount to field of use limitations described by 2106.05(h). These limitations do not describe how the model was trained only that is has been trained on certain data. Additionally, “receiving a language input in a source language, and (ii) data identifying a target task to be performed on the language input;” amounts to adding insignificant extra solution activity to the judicial exception, as receiving “named data” amounts to mere data gathering. The claim does not set forth any specifics on how the gathering is performed. All uses of the named machine learning models first require receiving data. See MPEP 2106.05(g).
The claim is directed to an abstract idea.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. Further, the additional elements, “receiving a language input in a source language, and (ii) data identifying a target task to be performed on the language input” are insignificant extra-solution activities that are considered well-understood, routine, conventional activities, for the following reason. Receiving named data amounts to receiving or transmitting data over a network (MPEP 2106.05(d)(II)(i). According to MPEP 2106.05(d)(II)(i), “The courts have recognized the following computer functions as well‐understood, routine, and conventional functions when they are claimed in a merely generic manner”. As such, the insignificant extra-solution activities are considered well-understood, routine, conventional activities. Therefore, the claim is not patent eligible
Regarding Claim 3/11/19
The claim is directed to a process/machine/article of manufacture. The claim recites the following limitations “wherein: the target task is a translation task that requires generating an output text in a target language; and the task identifier comprises a token identifying the target language.” Under Step 2A Prong 1, these limitations only serve to describe the abstract idea addressed in the independent claim.
Furthermore, under step 2A Prong 2 and 2B, the claim does not recite additional elements to consider other than those considered in the independent claim. Accordingly, the recited additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea, nor do they amount to significantly more than the judicial exception because they do not impose any meaningful limits on practicing the abstract idea.
Regarding Claim 4/12/20
The claim is directed to a process/machine/article of manufacture. The claim recites the following limitations Under Step 2A Prong 1, the claims do not describe any additional abstract ideas beyond those described in the parent claim.
Step 2A Prong Two Analysis: The judicial exception in not integrated into a practical application. In particular, the claims recite the additional element(s) the limitations “wherein the set of training examples comprise a plurality of paired datasets, wherein each of the paired datasets comprises an input dataset of text in a respective input language paired with an output dataset in a respective output language.” amounts to generally linking to a particular technological field of use. The parent claim describes that the machine learning model “has been trained”, the claim does not provide any detail on how the training is performed, see MPEP 2106.05(h).
Accordingly, the recited additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea, nor do they amount to significantly more than the judicial exception, under Step 2B, because they do not impose any meaningful limits on practicing the abstract idea.
Regarding Claim 5/13
The claim is directed to a process/machine. The claim recites the following limitations Under Step 2A Prong 1, the claims do not describe any additional abstract ideas beyond those described in the parent claim.
Step 2A Prong Two Analysis: The judicial exception in not integrated into a practical application. In particular, the claims recite the additional element(s) the limitations “wherein the plurality of paired datasets does not include a pairing of datasets comprising an input dataset in the source language paired with an output dataset in the target language.” amounts to generally linking to a particular technological field of use. The parent claim describes that the machine learning model “has been trained”, the claim does not provide any detail on how the training is performed, see MPEP 2106.05(h).
Accordingly, the recited additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea, nor do they amount to significantly more than the judicial exception, under Step 2B, because they do not impose any meaningful limits on practicing the abstract idea.
Regarding Claim 6/14
The claim is directed to a process/machine. The claim recites the following limitations “wherein the task identifier comprises a token identifying the target task.” Under Step 2A Prong 1, these limitations only serve to describe the abstract idea addressed in the independent claim.
Furthermore, under step 2A Prong 2 and 2B, the claim does not recite additional elements to consider other than those considered in the independent claim. Accordingly, the recited additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea, nor do they amount to significantly more than the judicial exception because they do not impose any meaningful limits on practicing the abstract idea.
Regarding Claim 7/15
The claim is directed to a process/machine. The claim recites the following limitations Under Step 2A Prong 1, the claims do not describe any additional abstract ideas beyond those described in the parent claim.
Step 2A Prong Two Analysis: The judicial exception in not integrated into a practical application. In particular, the claims recite the additional element(s) the limitations “wherein the decoder neural network has an attention mechanism.” amounts to generally linking to a particular technological field of use. The parent claim describes that the machine learning model “has been trained”, the claim does not provide any detail on how the training is performed, see MPEP 2106.05(h).
Accordingly, the recited additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea, nor do they amount to significantly more than the judicial exception, under Step 2B, because they do not impose any meaningful limits on practicing the abstract idea.
Regarding Claim 8/16
The claim is directed to a process/machine. The claim recites the following limitations Under Step 2A Prong 1, the claims do not describe any additional abstract ideas beyond those described in the parent claim.
Step 2A Prong Two Analysis: The judicial exception in not integrated into a practical application. In particular, the claims recite the additional element(s) the limitations “wherein the shared vocabulary is a shared word piece vocabulary that includes sub word units that are shared between the plurality of different candidate languages.” amounts to generally linking to a particular technological field of use. The parent claim describes that the machine learning model “has been trained”, the claim does not provide any detail on how the training is performed, see MPEP 2106.05(h).
Accordingly, the recited additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea, nor do they amount to significantly more than the judicial exception, under Step 2B, because they do not impose any meaningful limits on practicing the abstract idea.
Conclusion
Claim(s) 1, 3-9, 11-17, 19-20 have been searched, but have not been rejected with respect to 35 U.S.C. 102 nor 35 U.S.C. 103. Specifically, none of the reference of record either alone or in combination fairly disclose or suggest the limitation:
Generating a tokenized input by tokenizing the language input using a shared vocabulary that includes tokens representing sub-word units that are shared between a plurality of different candidate languagethe tokenized input and the selected task identifier comprises prepending the selected task identified identifying the target task to an output of the decoder neural network However, these claims remain rejected under 35 U.S.C. 101.
Prior art not relied upon: Sennrich et al “Linguistic Input Features Improve Neural Machine Translation” describes concatenating additional input features which may be interpreted as language identifiers as they vary depending on the target language.
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JOHNATHAN R GERMICK whose telephone number is (571)272-8363. The examiner can normally be reached M-F 9:30-4:30.
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, Kakali Chaki can be reached on 571-272-3719. 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.
/J.R.G./
Examiner, Art Unit 2122
/KAKALI CHAKI/ Supervisory Patent Examiner, Art Unit 2122
Prosecution Insights