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 .
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 1-15 are rejected under 35 U.S.C. 101 because the claimed invention is directed to abstract ideas without significantly more. The claims recite functional outcomes desired to be obtained from operating generic computers to train a generic machine learning model for a desired field of use (inferring controller operations from video frames). No technical details are provided for what hardware, software or system architecture are configured for accomplishing the claimed outcomes. The claims do not specify what type of machine learning model is used, how a particular machine learning model is trained, how it makes inferences, or how outputs are generated, for example. The claims are essentially an expression of a desire to train an existing machine learning model training in a conceived field of use. The claims lack any description of novel hardware and/or software or contain any evidence of how any particular ML model or the field of ML models are improved.
Claims to training a prior art machine learning model with certain types of data have been held to be ineligible in the Court of Appeals for the Federal Circuit in the suits against Fox Corp. Fox Broadcasting Company, LLC and Fox Sports Productions, LLC. Here the court considered that when claims merely use machine learning to claim an abstract idea itself by “using a generic machine learning technique in a particular environment,” this is a failure to transform the claimed abstract idea into “significantly more” – there is “no inventive concept.” The CAFC opinion concluded with a note that, “[m]achine learning is a burgeoning and increasingly important field and may lead to patent-eligible improvements in technology.” The court explained that its instant opinion held “only that patents that do no more than claim the application of generic machine learning to new data environments, without disclosing improvements to the machine learning models to be applied, are patent ineligible under § 101.”
The pending claims are drafted using functional language that lists desired end results (inputting, presenting) from operating computers of unclaimed technical specifications through unclaimed processing steps to train some unclaimed type of machine learning model in unclaimed manner(s) and generating outputs from the model in unspecified manner(s). Observing some existing video and inferring what controller operations might have gone into generating that video, absent any technical details of how computers are configured and programmed to do this, is rooted in human mental work comprising observations, evaluations, judgments, and formation of opinions, that belongs to the enumerated abstract idea grouping of “Mental Processes.” A human being observing video game scenes could conceive of possible game controller input sequences that a player might have made to accomplish game scenes viewed thereby. This mental process is longstanding in video gaming where a player wishes to learn from watching other players or ascertain how skillfully a player is performing.
Detailed Analysis
The following detailed analysis is based on the subject matter eligibility examination guidelines provided in the MPEP at https://www.uspto.gov/web/offices/pac/mpep/s2106.html
Steps 1 and 2 of the analysis have been conducted for all of the pending claims.
Step 1 (See MPEP §2106.03): In this step, it is determined whether the pending claims are directed to at least one of the four statutory categories of subject matter. Here it is determined that all of the pending claims fall into statutory categories. The claims meet step 1 as follows:
Claims 1-5 recite a process (method)
Claims 6-15 recite an article of manufacture (apparatus).
Step 2A, Prong 1 (see MPEP 2106.04(I)): In this step of the analysis, judicial exception(s) that fall into one or more of the abstract idea groupings enumerated in MPEP 2106.04(a) are identified.
The claims recite the following judicial exceptions:
“Mental Processes” (See MPEP § 2106.04(a)(2)(III)). Observing rendered video game scenes, evaluating and making judgments and forming opinions on what controller operations could have preceded or accompanied the scenes is activity traditionally performed by human beings – a notoriously well-known scenario being a spectator watching another person play a video game and thinking about what controller actions may have been used, either to learn how to play the game or to subjectively evaluate a player’s skill. The claims do not recite any particulars of how computers are programmed to process or analyze player inputs using a model or how machine learning models are used to generate outputs. The claims are essentially a list of functionally-recited desired end results for a conceived field of use for machine learning. The claims do not delineate steps through which any particular ML model achieves an improvement. Support for why the pending claimed invention is directed to abstract mental processes can be found in Electric Power Group v. Alstom, S.A., 830 F.3d 1350, 1353-54, 119 USPQ2d 1739, 1741-42 (Fed. Cir. 2016), wherein "collecting information, analyzing it, and displaying certain results of the collection and analysis," where the data analysis steps are recited at a high level of generality such that they could practically be performed in the human mind, were held to be abstract.
The following limitations of claim 1 are directed to judicial exceptions. Claim 1 is representative, independent claims 6 and 11 rise and fall with claim 1:
inputting to at least one machine learning (ML) model at least a training set…;
inputting to the ML model at least a first recorded computer simulation not including information about CSC operation…;
presenting the first recorded computer simulation along with information about CSC operations executed during generation of the first recorded computer simulation received from the ML model.
As discussed above, collecting information, analyzing it and displaying certain results of the collection and analysis, wherein the steps are recited at a high level of generality has been held to fall into the grouping of mental processes in Electric Power Group v. Alstom. The claims do not recite any details of computer programming steps or details of the type or use of a particular machine learning model or any tangible improvements to computers or to a machine learning model or the field of machine learning. As such the claimed ideas for collecting video frames and presenting information about the frames (wherein there are no details as to how the analysis occurs or what the content of presentation is or how this content is generated) are held to be abstract equivalents of human mental work.
Step 2A, Prong 2: In this step, any additional elements beyond the identified abstract ideas are identified and evaluated for any integration into a practical application. In particular, any claimed technological improvement is considered.
Additional elements recited in the claims include:
an apparatus comprising computer memory… (claim 1)
This hardware component and the implied-but-unclaimed software used to train a machine learning model are recited at a high level of generality and merely outline a generic technological field in which to apply the abstract ideas for applying generic machine learning to the practice of learning controller operations through video frames through routine and conventional use of generic hardware. The instant disclosure’s silence as to any of the claimed hardware or software having any nonobvious structural or functional specifications or requirements or being used to solve any stated problems existing in computers per se or to provide any improvements to computers per se supports a finding that the hardware additional elements of the claimed apparatus comprises merely generic components and technologies used in their routine and conventional capacities in the art. “[T]he invocation of ‘already-available computers that are not themselves plausibly asserted to be an advance … amounts to a recitation of what is well-understood, routine, and conventional.” Customedia Techs., LLC v. Dish Network Corp., 951 F.3d 1359, 1366 (Fed. Cir. 2020). And “simply adding a general-purpose computer or computer components after the fact to an abstract idea […] does not integrate a judicial exception into a practical application or provide significantly more.” Affinity Labs v. DirectTV, 838 F.3d 1253, 1262, 120 USPQ2d 1201, 1207 (Fed. Cir. 2016)
intended manners in which analyzed data is to be presented-- audibly, visibly, as well as including some data associations having unclaimed particulars. (Dependent claims 2-5, 7-10 and 12-15)
These claims recite, functionally and without any details of computer programming steps or required hardware/software configurations, the form analyzed data takes in some unspecified type of output presentation. There is an absence of any tangible evidence of improvements to computers per se or to the technical field of machine learning, and as such the form data resulting from analysis takes is a matter of insignificant extra-solution activity. As explained by the Supreme Court, the addition of insignificant extra-solution activity does not amount to an inventive concept, particularly when the activity is well-understood or conventional. Parker v. Flook, 437 U.S. 584, 588-89, 198 USPQ 193, 196 (1978). In Flook, the Court reasoned that “[t]he notion that post-solution activity, no matter how conventional or obvious in itself, can transform an unpatentable principle into a patentable process exalts form over substance.”
Claims 2-5, 70-10 and 12-15 are also seen as “the use of a computer or other machinery in its ordinary capacity for economic or other tasks (e.g., to receive, store, or transmit data) or simply adding a general-purpose computer or computer components after the fact to an abstract idea […] does not integrate a judicial exception into a practical application or provide significantly more.” Affinity Labs v. DirectTV, 838 F.3d 1253, 1262, 120 USPQ2d 1201, 1207 (Fed. Cir. 2016 attributing particular generic computer functions for computer hardware to perform from well-known, routine, conventional functions performed by such hardware has been held to be insufficient to show an improvement to technology, Affinity Labs of Tex. v. DirecTV, LLC, 838 F.3d 1253, 1264, 120 USPQ2d 1201, 1208 (Fed. Cir. 2016).
The preceding additional elements, considered alone and in the context of the claims, do not integrate the abstract game management or game rules into a practical application that improves computer functionality or another technology. They:
Invoke generic computers, memories, and conventional networked game environments. See where the instant specification discloses a standard CPU, input, display, storage, server, etc. (e.g., on p. 2 of the specification, a computer simulation console processor or processor embedded in a display, and a display capable of presenting video effected by a game controller). “[T]he invocation of ‘already-available computers that are not themselves plausibly asserted to be an advance … amounts to a recitation of what is well-understood, routine, and conventional.” Customedia Techs., LLC v. Dish Network Corp., 951 F.3d 1359, 1366 (Fed. Cir. 2020). And “simply adding a general-purpose computer or computer components after the fact to an abstract idea […] does not integrate a judicial exception into a practical application or provide significantly more.” Affinity Labs v. DirectTV, 838 F.3d 1253, 1262, 120 USPQ2d 1201, 1207 (Fed. Cir. 2016)
Do not recite a specific improvement to the functioning of a computer (e.g., no improved rendering pipeline, no reduced latency synchronization protocol, no novel memory management, no graphics or physics engine enhancement).
Do not effect a transformation of an article.
Are drafted as applying the abstract idea in the field of computer games (field-of-use) with result-oriented language (e.g., “inputting”, “presenting”).
With regard to interpreting result-oriented claim language when performing a 35 USC §101 analysis, see Beteiro LLC v. DraftKings Inc., (Fed. Cir 2024) when "the claims are drafted using largely (if not entirely) result-focused functional language, containing no specificity about how the purported invention achieves these results. Claims of this nature are almost always found to be ineligible for patenting under Section 101." See also Interval Licensing LLC v. AOL Inc. (896 F.3d 1335) wherein the court found that claims to a computer software "attention manager" that displays content on unused portions of a screen were result-oriented and invalid under 35 U.S.C. § 101 because they did not recite a specific technological method for achieving the claimed result; Contour IP Holding LLC v. GoPro, Inc., 2024 U.S. App. LEXIS 22825 (Fed. Cir. 2024): The court held that claims must not only describe desired outcomes but also include a specific process or machinery for achieving that result; In re Killian, 45 F.4th 1373 (Fed. Cir. 2022): The court reaffirmed that claims simply reciting a desired result without specifying how to achieve it are directed to an abstract idea and are ineligible under 35 U.S.C. § 101. The claims at issue were directed to analyzing data from two databases. In the Step Two of the Alice test, the court determined that there was no inventive concept because the additional elements merely involved generic and routine data gathering and analysis steps that could have been performed with or without a computer.
MPEP § 2106.05(f) explains that, “The recitation of claim limitations that attempt to cover any solution to an identified problem with no description of the mechanism for accomplishing the result, does not integrate a judicial exception into a practical application or provide significantly more because this type of recitation is equivalent to the words "apply it"”.
The pending claims do not include any technical description of mechanisms for accomplishing the claimed results. Instead, the claims use some unspecified computer and unspecified programming to conduct generic, result-oriented steps such as “inputting”, “presenting”, “audibly presenting” and “visibly presenting”, for performing equivalents of mental processes. The claims seek to cover any system and any method (such as any hardware devices, any programming instructions) for applying the abstract machine learning model training As such the claims are found to be directed to ineligible subject matter.
Step 2A Prong 2 concludes in a determination that the additional elements do not amount to a practical application of the claimed abstract ideas.
Step 2B: In this step of the Alice analysis, it is assessed whether additional elements amount to significantly more than abstract ideas. Any well-understood, routine, conventional (“WURC”) activity is also discussed along with evidentiary considerations.
Absent integration into a practical application, the claims lack “significantly more” than the abstract idea.
Additional elements that are generic computer implementation and conventional components are:
“computer memory”, “a processor”
The specification characterizes these computing components as conventional computing hardware and software performing ordinary functions (spec. p. 2, supporting a finding that the implementation is well-understood, routine, and conventional (WURC). See Berkheimer v. HP Inc., 881 F.3d 1360 (Fed. Cir. 2018) (WURC must be supported); here, the instant specification itself indicates conventionality.
Conclusion:
Claims 1-15 are found to be ineligible under 35 U.S.C. § 101. Although step 1 is satisfied (the claims recite manufacture/process/machine), in Step 2A Prong 1, the claims are found to recite an abstract idea—training a generic machine learning model to learn controller operations based on observed video game frames, wherein this is activity rooted in human mental observation, opinion and judgement. And as found in Step 2A Prong 2, the abstract ideas are not integrated into a practical application; only generic computer implementation and field-of-use limitations are claimed. There are no technical details in the claims that reveal how any of the claimed result-oriented language is to be accomplished. And performing Step 2B, there is nothing “significantly more” found beyond WURC elements as evidenced by the specification.
Possible remedies:
To improve subject matter eligibility under 35 USC § 101, it is recommended to anchor the claims to concrete, non-generic technical mechanisms (such as particular software processes or nonobvious system architectures) in a way that there is evidence in the claims of certain improvements to computer or network operations or to another technology.
In the field of the instant invention (training a machine learning model), an improvement would have to be found to an inherently technical problem existing in computers and would have to reveal how the computer(s) themselves or the field of machine learning are improved as a direct result of the claimed invention. The details of the improvement to computers cannot be found in the wording of the abstract ideas (details of data collection, analysis or outputting that are equivalents of pre-digital era human mental work) themselves. Genetic Techs v Merial, an inventive concept "cannot be furnished by the unpatentable law of nature" itself. A subjective improvement in a game player’s user experience (by providing a game that might provide arguably unique rules) is not an improvement to computers themselves or to computer technology and does not solve any stated problem that is inherently technical in nature.The court ruled in International Business Machines Corporation v. Zillow Group, Inc., (CAFC, 17 October, 2022), that "improving a user's experience while using a computer application is not, without more, sufficient to render the claims" patent-eligible. Customedia Techs., LLC v. Dish Network Corp., 951 F.3d 1359, 1365 (Fed. Cir. 2020).
Examples might include to:
Tie abstract steps to a specific, non-generic technological implementation that improves computer functionality or another technology in some tangible way (e.g., reduces network latency by X, improves memory utilization via Y, improves image fidelity through Z), with technical mechanisms claimed.
Provide evidence of improvements to computers or network operations in the claims by claiming certain network nonobvious server-side architecture that is also claimed as solving problems existing in the art, or claiming a certain improvement in rendering such as a GPU-accelerated improvement that provides measurable improvements to game functionality.
Add claim elements showing a particular machine or a transformation of an article, beyond mere data manipulation or display functions.
Replace result-oriented terms (“execute…”, “control a plurality of types of objects,” “cause a fourth object to appear”) with concrete steps and parameters tied to the technical mechanism (e.g., explicit algorithmic operations, message formats, timing constraints, thresholds).
Limit scope to a specific technological field and architecture (e.g., “a distributed game server cluster employing [named protocol] with defined message cadence and buffer management”) and claim the architecture itself, avoiding broad “apply it on a computer” formulations.
Provide specification support demonstrating the asserted improvements are not well-understood, routine and conventional:
Implementation details: algorithms with stepwise operations, data structures with constraints, hardware configurations, protocol diagrams.
Performance evidence: benchmarks, latency/throughput graphs, memory usage comparisons versus baselines.
Engineering rationale: why existing approaches fail and how your mechanism achieves measurable gains.
Recite in the claims a technical solution to a technological problem (e.g., secure hardware-backed attestations, novel protocol flows, improved cryptographic operations, sensor fusion pipelines).
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.
Claims 1-15 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by US 2020/0269136 A1 to Gurumurthy et al.
Re claims 1, 6 and 11, Gurumurthy teaches a method and an apparatus comprising: computer memory comprising instructions executable by at least one processor system comprising one or more processors,
[0037] discloses that the invention of the disclosure is carried out using processors 312, 318, or 334 including a central processor (CPU), or by GPU’s. [0073] describes an illustrative computing device 800 that comprises at least one processor 802 for executing instructions that can be stored in memory 804. This computer can be communicatively coupled with other computers over a network such as the Internet. [0074] describes that client devices of the invention of the disclosure can be general purpose personal computers, cell phones or other mobile devices.
the instructions being executable to: input to at least one machine learning (ML) model at least a training set, the training set comprising sequences of video frames from plural recorded computer simulations and information associated with the sequences of video frames about computer simulation controller (CSC) operations executed during generation of the sequences of video frames; input to the ML model at least a first recorded computer simulation not including information about CSC operations executed during generation of the first recorded computer simulation;
The abstract describes that the invention of Gurumurthy provides "personalized coaching" to "players of an electronic gaming application" wherein "Data can be obtained that demonstrates how skilled players" such as "professional players play a game". "This data can be used to train a machine learning model for the game [...] The model can infer one or more actions [...] The information can be conveyed to the player using visual, audio, or haptic guidance during gameplay, or can be provided offline, such as with video or rendered replay of the game session." Refer also to Figs. 2A-2F for illustrations of some AI coaching overlays on video game video frames that provide indications of control operations that should be taken by a player, wherein these control operations were learned by a machine learning model observing game play by a skilled or professional player as will be discussed in further detail below.)
[0017] describes that "neural networks can be used to infer various actions" which can be beneficial to output "displaying information [...] indicating an action the player should make in the game"
[0029] describes to, "provide a machine learning—or artificial intelligence-based virtual coach, which can assist players, such as e-sports gamers, in learning and/or improving their gameplay for at least certain games […] such as Fortnite or Counter-Strike, which attract a large number of players of different skill levels, from professional or other highly-skilled players to new or novice players who may not know even the basics of gameplay.”
[0032] describes that the invention of the disclosure, "can capture player data as discussed herein, then apply deep learning to determine effective actions and strategies for a particular game or type of game. This can include, for example, obtaining image or video data for a game session and using computer vision to analyze the individual images for video frames to determine actions being taken, as well as the current state of the game. [...] This can be performed for data from a number of experienced or professional players, for example, such that effective strategies can be learned based on what are likely to be the most successful sessions of that game."
[0038] describes that “training data” for a machine learning virtual coach may comprise video demos of gameplay wherein “The files may be video replay files (from Twitch, YouTube, etc.) of professional human players that can be analyzed and the inputs or actions determined or inferred … In embodiments where the gameplay data is not available through an exposed API [...] player inputs at specific states of gameplay, or points in a gameplay session [...] can be determined by analyzing the image or video data in at least some embodiments. For example, image data can be captured for frame-by-frame analysis […]”
and present the first recorded computer simulation along with information about CSC operations executed during generation of the first recorded computer simulation received from the ML model
[0018] describes that a game used with the invention of the disclosure is a "three-dimensional (3D) first-person shooter-type game" that is played wherein "The player can thus provide input, such as by tapping keys of a keyboard or pressing buttons of a joypad controller [...] to move through the world [...] to switch or fire weapons, run, crouch, jump, etc. [...] This can make it difficult for many novice players to quickly get up to speed with the game, as the players must not only learn the strategy of the game and figure out what to do, but must also attempt to learn the specific inputs and combinations that can trigger the desired actions [...] the amount that the player has to learn and remember can become overwhelming"
[0020] describing Fig. 1 indicates that, “the state of the game is analyzed [...] a series of actions can be determined, and these can be conveyed in a reasonable sequence of actions to be taken [...]”
[0032] describes that the invention, "can capture player data as discussed herein, then apply deep learning to determine effective actions and strategies for a particular game or type of game. This can include, for example, obtaining image or video data for a game session and using computer vision to analyze the individual images for video frames to determine actions being taken [...] This can be performed for data from a number of experienced or professional players”
[0038] describes that, "the files may be video replay files (from Twitch, YouTube, etc.) of professional human players that can be analyzed and the inputs or actions determined or inferred [...] a stream of player input can be obtained and parsed for the relevant information [...] player inputs at specific states of gameplay, or points in a gameplay session [...] can be determined by analyzing the image or video data […] image data can be captured for frame-by-frame analysis of a gameplay session for a specific player session, and player input can be determined that can be correlated for each frame, in order to recreate the state of the game and player actions. This data can then be used as training data when obtained for the highly-skilled player, and used to determine coaching advice when determined for lesser-skilled players or players otherwise taking advantage of a virtual coach as discussed herein."
and
present the recorded computer simulation on at least one audio video (AV) display along with at least one indication of at least one of the CSC operations received from the ML model.
[0012] describes that "data can be obtained that demonstrates how [..] professional players play a specific game. […] Gameplay data [...] can be obtained [...] by analyzing game-related information such as displayed content." and wherein "The information can then be conveyed to the player in a way that best helps the player. This can include, for example, providing visual [...] guidance in near real time during gameplay. For example, advice can be provided as to the recommended next action for the player to take in the game. The advice can also be provided […] such as with video or rendered playback or replay of the game session."
[0020] describes that, “[...] a series of actions can be determined, and these can be conveyed in a reasonable sequence of actions to be taken [...] the game might coach the player to move the avatar 102 to that location 104, then ready a weapon, lean around the corner, and fire at the gameplay element 106. [...]
[0021], "FIGS. 2A through 2F illustrate examples of advice or guidance that a virtual coach might provide to a gamer in accordance with various embodiments. [...] In the example image 200 of FIG. 2A, a graphical overlay 202 is provided indicating that it has been determined to be advantageous for the user to consider switching to a different weapon, or equipping a different item. In this example, the advice is for the user to switch from a pistol to a grenade before progressing further along the current path. The overlay can take any of a number of different forms, including text, an image of the grenade, etc. [...] If it is urgent that the player switch to a different item then the item might glow red or flash quickly, etc. For novice players, the advice may include instructions on switching to the grenade, such as the next key or button to press to take that action"
[0022], "FIG. 2B illustrates another example image 210 of advice that can be provided in accordance with various embodiments. In this example, the advice indicates a path for the player to take to be more likely to obtain the determined goal. This may include providing indication of a preferred path 212 based on prior gameplay of other players."
Re claims 2-3, 7-8, 12-13, in addition to conveying suggested video game control commands inferred from machine learning frame by frame image analysis of professional players using graphics such as overlays on game play, Gurumurthy additionally teaches audio or haptic guidance can be used to convey information to users during gameplay. See the abstract, [0012], [0019], [0030], “the virtual agent can provide “spoken” communications to the human player as well, such as to tell the player actions to take, a strategy to follow”.
Re claims 4-5, 9-10. 14-15, these claims recite that computer games used to train an ML model to infer controller actions could have been executed locally or at a server. See [0074] which discloses that client terminals can either perform their own processing or can be thin-client terminals where their processing is performed over a network. [0012] describes that gameplay data can be obtained directly from a game or from a game server. And [0034] describes that games played by professional players that are used as training data for the ML coach described above can either be “hosted locally … or on a number of game servers.”
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to STEVEN J HYLINSKI whose telephone number is (571)270-1995. The examiner can normally be reached Mon-Fri 10-530.
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, Dmitry Suhol can be reached at (571) 272-4430. 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.
/STEVEN J HYLINSKI/Primary Examiner, Art Unit 3715
Prosecution Insights