DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 16 December 2024, 10 February 2026, and 30 June 2026, respectively, are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
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, 6-7, 12, and 14-15 is/are rejected under 35 U.S.C. 102(a)(2) as being anticipated by “AutoMix: Automatically Mixing Language Models”, hereinafter referred to as Mandaan et al.
Regarding claim 1, Mandaan et al. discloses a computer-implemented method comprising:
receiving an input query from a user (“The query first goes to the SLM, and an AutoMixSLM−MLM decides between reporting the SLM answer or routing to the MLM,” Mandaan et al., p. 8, highlighted sentence.);
evaluating the input query based, at least in part, on a first language model (Mandaan et al., p. 8, highlighted sentence. Here, the first language model is the SLM.);
selecting a second language model from a plurality of second language models based, at least in part, on evaluating the input query (Mandaan et al., p. 8, highlighted sentence. Here, the second language model is the MLM or LLM.);
processing the input query based, at least in part, on the second language model (“The preceding discussion focused on a two-model scenario involving the SLM and LLM. This section extends this framework to incorporate a third model, the MLM. Our decision flow commences with the SLM generating an answer, which is then self-verified by the SLM. The verifier probability serves as an observation, guiding one of the following actions: 1) Reporting the SLM answer, 2) Running inference on the MLM or LLM and reporting the answer, or 3) Running inference on the MLM and verifying the answer,” Madaan et al., p. 4, highlighted sentence. Here, the input query is processed by the second language model (i.e., MLM or LLM).);
performing an operation based, at least in part, on the processing of the input query (Madaan et al., p. 4, highlighted sentence. Here, the operation is verifying the answer.).
As to claim 14, system claim 14 and method claim 1 are related as method and system of using same, with each claimed element’s function corresponding to the method step. Accordingly claim 14 is similarly rejected under the same rationale as applied above with respect to method claim. And, the experimental results described in Mandaan et al., section 4, indicates storage (CRM), processor(s), and instructions.
Regarding claim 6, Mandaan et al. discloses the computer-implemented method according to claim 1,
wherein evaluating the input query comprises a complexity evaluation of the input query (“In the context of meta-verifier, we observe that all the queries in this two language model setup could be categorized in three different categories: Simple, Complex, and Unsolvable,” Mandaan et al., p. 3, highlighted section.); and
wherein selecting the second language model is based on the complexity evaluation of the input query (Madaan et al., p. 3, highlighted section.).
Regarding claim 7, Mandaan et al. discloses the computer-implemented method according to claim 1,
wherein the second plurality of language models comprises models of different complexity (Mandaan et al., p. 3, highlighted section.); and
selecting the second language model is based on the complexity of the second language model (Madaan et al., p. 3, highlighted section.).
Regarding claim 12, Mandaan et al. discloses the computer-implemented method according to claim 1, wherein the first language model differs from each of the plurality of second language models by: a number of parameters of the model (Mandaan et al., p. 8, highlighted sentence. An SLM has fewer parameters than an MLM or LLM.), a dimension of the input of the model, an average number of bitwise operations for processing one input query, an average number of CPU cycles for processing one input query, and/or a number of samples used for training the model (Mandaan et al., p. 8, highlighted sentence. An SLM typically uses fewer training samples than an MLM or LLM.).
Regarding claim 15, Mandaan et al. discloses a car comprising at least one processing unit configured to perform a method including:
receiving an input query from a user (“The query first goes to the SLM, and an AutoMixSLM−MLM decides between reporting the SLM answer or routing to the MLM,” Mandaan et al., p. 8, highlighted sentence.);
evaluating the input query based, at least in part, on a first language model (Mandaan et al., p. 8, highlighted sentence. Here, the first language model is the SLM.);
selecting a second language model from a plurality of second language models based, at least in part, on evaluating the input query (Mandaan et al., p. 8, highlighted sentence. Here, the second language model is the MLM or LLM.);
processing the input query based, at least in part, on the second language model (“The preceding discussion focused on a two-model scenario involving the SLM and LLM. This section extends this framework to incorporate a third model, the MLM. Our decision flow commences with the SLM generating an answer, which is then self-verified by the SLM. The verifier probability serves as an observation, guiding one of the following actions: 1) Reporting the SLM answer, 2) Running inference on the MLM or LLM and reporting the answer, or 3) Running inference on the MLM and verifying the answer,” Madaan et al., p. 4, highlighted sentence. Here, the input query is processed by the second language model (i.e., MLM or LLM).);
performing an operation based, at least in part, on the processing of the input query (Madaan et al., p. 4, highlighted sentence. Here, the operation is verifying the answer.) and further having:
an input system configured to obtain a user input corresponding to an input query (Madaan et al., fig. 1 – AutoMix is a system which takes an input query.);
at least one element controllable by the system in response to an input query (Madaan et al., fig. 1 – Based on the meta-verifier’s decision, either the initial answer (1890 AD) is returned, or the question is rerouted to a larger language model (LLM) to enhance accuracy. The examiner notes that rerouting is controlling the language models of the (AutoMix) system.).
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.
Claim(s) 2-5 and 16-18 is/are rejected under 35 U.S.C. 103 as being unpatentable over “AutoMix: Automatically Mixing Language Models”, hereinafter referred to as Mandaan et al., in view of US 20160307568, hereinafter referred to as Zhao et al.
Regarding claim 2, Mandaan et al. discloses the computer-implemented method according to claim 1, but not wherein evaluating the input query comprises selecting one or more applications from a plurality of applications, wherein the plurality of applications includes at least one vehicle-related application and/or at least one navigation-related application. Zhao et al. is cited to disclose wherein evaluating the input query comprises selecting one or more applications from a plurality of applications, wherein the plurality of applications includes at least one vehicle-related application (“Telematics unit 30 can be used to provide a diverse range of vehicle services that involve wireless communication to and/or from the vehicle. Such services include: turn-by-turn directions and other navigation-related services that are provided in conjunction with the GPS-based vehicle navigation module 40,” Zhao et al., para [0018].) and/or at least one navigation-related application (“Telematics unit 30 can be used to provide a diverse range of vehicle services that involve wireless communication to and/or from the vehicle. Such services include: turn-by-turn directions and other navigation-related services that are provided in conjunction with the GPS-based vehicle navigation module 40,” Zhao et al., para [0018].). Zhao et al. benefits Mandaan et al. by identifying vehicle functions to be controlled by a gate command that is associated with it, thereby improving the ASR system speed and accuracy (Zhao et al., Background). Therefore, it would be obvious for one skilled in the art to combine the teachings of Mandaan et al. with those of Zhao et al. to extend the language model routing application of Mandaan et al.
Regarding claim 3, Mandaan et al., as modified by Zhao et al., discloses the computer-implemented method according to claim 2, wherein the plurality of applications comprises one or more of: a routing application (“The navigation services can be provided using a dedicated in-vehicle navigation module (which can be part of GPS module 40), or some or all navigation services can be done via telematics unit 30, wherein the position information is sent to a remote location for purposes of providing the vehicle with navigation maps, map annotations (points of interest, restaurants, etc.), route calculations, and the like,” Zhao et al., para [0019].), a search application, electrical vehicle (EV) charger search application, a traffic application, a vehicle control application (“The system and method described below receives speech from a vehicle occupant that includes commands controlling one or more vehicle functions,” Zhao et al., para [0010].), a weather application, a web search application, a general knowledge application, a feedback application, a car manual application, and a Heating, Ventilation, and Air Conditioning (HVAC) control application.
Regarding claim 4, Mandaan et al., as modified by Zhao et al., discloses the computer-implemented method according to claim 2, wherein performing the operation is performed using, at least in part, the one or more applications (Zhao et al., para [0018].).
Regarding claim 5, Mandaan et al., as modified by Zhao et al., discloses the computer-implemented method of claim 4, wherein using the application further comprises:
transmitting a second input query to a first server (“Similarly, speech recognition software can be processed using processors of one of the servers 82 in the call center 20. In other words, the ASR system 210 can be resident in the telematics unit 30, distributed across the call center 20 and the vehicle 12 in any desired manner, and/or resident at the call center 20,” Zhao et al., para [0031].); and
receiving a processed second input query from the first server (Zhao et al., para [0040] – post-processor 216.).
Regarding claim 16, Mandaan et al., as modified by Zhao et al., discloses the computer-implemented method according to claim 3, wherein performing the operation is performed using, at least in part, the one or more applications (“As used herein, the term ‘vehicle user interface’ broadly includes any suitable form of electronic device, including both hardware and software components, which is located on the vehicle and enables a vehicle user to communicate with or through a component of the vehicle,” Zhao et al., para [0021]. The software components allowing a user to interact with the vehicle system to perform specific tasks are applications.).
Regarding claim 17, Mandaan et al., as modified by Zhao et al., discloses the computer-implemented method according to claim 2,
wherein evaluating the input query comprises a complexity evaluation of the input query (“In the context of meta-verifier, we observe that all the queries in this two language model setup could be categorized in three different categories: Simple, Complex, and Unsolvable,” Mandaan et al., p. 3, highlighted section.); and
wherein selecting the second language model is based on the complexity evaluation of the input query (Madaan et al., p. 3, highlighted section.).
Regarding claim 18, Mandaan et al., as modified by Zhao et al., discloses the computer-implemented method according to claim 2,
wherein the second plurality of language models comprises models of different complexity (Mandaan et al., p. 3, highlighted section.); and
selecting the second language model is based on the complexity of the second language model (Mandaan et al., p. 3, highlighted section.).
Claim(s) 8-9 is/are rejected under 35 U.S.C. 103 as being unpatentable over “AutoMix: Automatically Mixing Language Models”, hereinafter referred to as Mandaan et al., in view of US 20190115017, hereinafter referred to as Sim.
Regarding claim 8, Madaan et al. discloses the computer-implemented method according to claim 1, but not wherein processing the input query comprises: transmitting the input query to a second server; receiving a processed input query from the second server; and controlling at least a part of a machine in response to the input query.
Sim is cited to disclose wherein processing the input query comprises:
transmitting the input query to a second server (“The speech recognition based vehicle control method may further include transmitting the speech command to the second server when the speech command is determined to be valid so that control corresponding to the speech command is performed,” Sim, para [0022]. The speech command is the input query.);
receiving a processed input query from the second server (Sim, para [0022], describes control corresponding to the speech command being performed (i.e., processing an input query).); and
controlling at least a part of a machine in response to the input query (Sim, para [0022], describes vehicle control in response to an input query.). Sim benefits Madaan et al. by providing a speech recognition-based vehicle control method in which user authentication is privately performed through a portable device (e.g., a wearable device) of a user without directly speaking information for authentication (e.g., a password or a personal identification number (PIN) code) in a speech form so that security when the user authentication is performed is sufficiently ensured. Therefore, it would be obvious for one skilled in the art to combine the teachings of Mandaan et al. with those of Sim to incorporate and simplify user authentication in the system described by Mandaan et al.
Regarding claim 9, Madaan et al., as modified by Sim, discloses the computer-implemented method according to claim 8, wherein the machine is a vehicle and controlling the machine preferably comprises one or more of: turning a wiper on or off, adjusting the ventilation, adjusting the heating (“For example, in a case in which a temperature of a seat of the vehicle 100 is to be increased in advance, when the user 110 says “turn on the hot-wire seat of the vehicle,” the home speaker 130 transmits the speech command of the user 110 to a telematics server 160 through an Internet of things (IoT) server 150 to control the vehicle 100,” Sim, para [0036].), closing or opening a window, controlling one or more vehicle lights, adjusting the position of a mirror, and closing or opening a sunroof.
Claim(s) 10 is/are rejected under 35 U.S.C. 103 as being unpatentable over “AutoMix: Automatically Mixing Language Models”, hereinafter referred to as Mandaan et al., in view of US 20240419745, hereinafter referred to as Thomson et al.
Regarding claim 10, Mandaan et al. discloses the computer-implemented method according to claim 1, but not wherein evaluating the input query comprises translation from a first language to a second language. Thomson et al. is cited to disclose wherein evaluating the input query comprises translation from a first language to a second language (“For example, a group of one or more transcription units may provide multiple services such as:…10. language translation,” Thomson et al., para [0778]-[0788].). Thomson et al. benefits Mandaan et al. by including transcription of ASR during an audio communication session, thereby assisting people that are hard-of-hearing to participate in the audio communications (Thomson et al., Background). Therefore, it would be obvious for one skilled in the art to combine the teachings of Mandaan et al. with those of Thomson et al. to extend the communication capabilities of by Mandaan et al.
Claim(s) 11 is/are rejected under 35 U.S.C. 103 as being unpatentable over “AutoMix: Automatically Mixing Language Models”, hereinafter referred to as Mandaan et al., in view of US 20250210033, hereinafter referred to as Sharifi et al.
Regarding claim 11, Madaan et al. discloses the computer-implemented method according to claim 1, but not wherein evaluating the input query is based on previous communication.
Sharifi et al. is cited to disclose wherein evaluating the input query is based on previous communication (“For instance, the metadata 254 may include an activity (e.g., driving, walking, etc.) the user 10 was performing when the user input the refinement query 250, a modality for how the user 10 input the refinement query 250 (e.g., through spoken input or typed input), a type of user device (e.g., smart phone, desktop/laptop, tablet, smart speaker, vehicle infotainment, etc.) the user 10 was using to interact with the assistant LLM when the refinement query 250 was input, or any other contextual information such as a history of the previous conversation in which the refinement query 250 was input by the user ,” Sharifi et al., para [0035].) and/or the plurality of the applications. Sharifi et al. benefits Mandaan et al. by providing a query replay technique that automatically applies a preferred refinement query that is relevant to a natural language query input by the user to the assistant LLM to cause the assistant LLM to generate a response to the current query that is personalized for the user (Sharifi et al., para [0028]). Therefore, it would be obvious for one skilled in the art to combine the teachings of Mandaan et al. with those of Sharifi et al. to allow the system of Mandaan et al. to make a clear decision about which user preferences are permanent, which preferences may change over time, and which preferences are context dependent.
Claim(s) 13 is/are rejected under 35 U.S.C. 103 as being unpatentable over “AutoMix: Automatically Mixing Language Models”, hereinafter referred to as Mandaan et al., in view of US 20160307568, hereinafter referred to as Zhao et al., and further in view of US 11527247 , hereinafter referred to as Kim.
Regarding claim 13, Madaan et al. discloses the computer-implemented method according to claim 2, but not wherein evaluating the input query comprises determining that at least one selected application is not available; or wherein processing the input query comprises determining that at least one selected application is not available.
Kim is cited to disclose wherein evaluating the input query comprises determining that at least one selected application is not available (“When the requested function or software may not be currently obtained from the function/service provider server 1600 according to a search result but information indicating when the function or software is available or updatable is received therefrom, the speech recognition agent 211 may store the information in a dialogue system of the speech analysis module 222 and provide the information when the user requests the function or service later. For example, when a request for a service currently unavailable in the proposed voice assistant is received from the user, instead of a response such as “The requested function is not currently serviced”, the proposed voice assistant may provide a notification about when the service requested by the user is available, e.g., “The requested service will be available next month. Thanks for understanding”, to the user and thus user convenience may be increased by providing information more than the notification merely indicating that the service is unavailable, to allow the user to expect the service,” Kim, col. 34, line 63 – col. 35, line 14.); or
wherein processing the input query comprises determining that at least one selected application is not available (“When the requested function or software may not be currently obtained from the function/service provider server 1600 according to a search result but information indicating when the function or software is available or updatable is received therefrom, the speech recognition agent 211 may store the information in a dialogue system of the speech analysis module 222 and provide the information when the user requests the function or service later. For example, when a request for a service currently unavailable in the proposed voice assistant is received from the user, instead of a response such as “The requested function is not currently serviced”, the proposed voice assistant may provide a notification about when the service requested by the user is available, e.g., “The requested service will be available next month. Thanks for understanding”, to the user and thus user convenience may be increased by providing information more than the notification merely indicating that the service is unavailable, to allow the user to expect the service,” Kim, col. 34, line 63 – col. 35, line 14.). Kim benefits Mandaan et al. by alerting a user to the unavailability of an ASR application, as well as when it will be available. Therefore, it would be obvious for one skilled in the art to combine the teachings of Mandaan et al. with those of Kim to enhance the user experience of the language model system as described by Mandaan et al.
Claim(s) 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over “AutoMix: Automatically Mixing Language Models”, hereinafter referred to as Mandaan et al., in view of US 20160307568, hereinafter referred to as Zhao et al., and further in view of US 20190115017, hereinafter referred to as Sim.
Regarding claim 19, Madaan et al., as modified by Zhao et al., discloses the computer-implemented method according to claim 2, but not wherein processing the input query comprises: transmitting the input query to a second server; receiving a processed input query from the second server; and controlling at least a part of a machine in response to the input query.
Sim is cited to disclose wherein processing the input query comprises:
transmitting the input query to a second server (“The speech recognition based vehicle control method may further include transmitting the speech command to the second server when the speech command is determined to be valid so that control corresponding to the speech command is performed,” Sim, para [0022]. The speech command is the input query.);
receiving a processed input query from the second server (Sim, para [0022], describes control corresponding to the speech command being performed (i.e., processing an input query).); and
controlling at least a part of a machine in response to the input query (Sim, para [0022], describes vehicle control in response to an input query.). Sim benefits Madaan et al. by providing a speech recognition-based vehicle control method in which user authentication is privately performed through a portable device (e.g., a wearable device) of a user without directly speaking information for authentication (e.g., a password or a personal identification number (PIN) code) in a speech form so that security when the user authentication is performed is sufficiently ensured. Therefore, it would be obvious for one skilled in the art to combine the teachings of Mandaan et al. with those of Sim to incorporate and simplify user authentication in the system described by Mandaan et al.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ANNE L THOMAS-HOMESCU whose telephone number is (571)272-0899. The examiner can normally be reached Mon-Fri 8-6.
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, Bhavesh M Mehta can be reached on 5712727453. 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.
/ANNE L THOMAS-HOMESCU/Primary Examiner, Art Unit 2656
Prosecution Insights