This specification provides a computer-implemented method, the method comprising obtaining a machine-learning model. The machine-learning model is being trained with expert data comprising a plurality of training examples. Each training example comprises: (i) game state data representing a state of a video game environment, and (ii) scored action data representing an action and a score for that action if performed by a video game entity of the video game environment subsequent to the state of the video game environment. An action is performed by the video game entity based on a prediction for the action generated by the machine-learning model. The method further comprises determining whether the action performed by the video game entity was optimal. In response to determining that the action performed by the video game entity was suboptimal, a healed training example is generated. The healed training example comprises: (i) the state of the instance of the video game environment, and (ii) healed scored action data indicative that the action performed by the video game entity was suboptimal. The machine-learning model is updated based on the healed training example.

This specification provides a computer-implemented method, the method comprising obtaining a machine-learning model. The machine-learning model is being trained with expert data comprising a plurality of training examples. Each training example comprises: (i) game state data representing a state of a video game environment, and (ii) scored action data representing an action and a score for that action if performed by a video game entity of the video game environment subsequent to the state of the video game environment. An action is performed by the video game entity based on a prediction for the action generated by the machine-learning model. The method further comprises determining whether the action performed by the video game entity was optimal. In response to determining that the action performed by the video game entity was suboptimal, a healed training example is generated. The healed training example comprises: (i) the state of the instance of the video game environment, and (ii) healed scored action data indicative that the action performed by the video game entity was suboptimal. The machine-learning model is updated based on the healed training example.

A mobile and adaptable fitness system allows a fitness device to be used to control an application running on a host device. The system includes one or more controllers, a motion sensing device, and a host device. The hand controllers may be mounted on the handlebars of the fitness device or held in a user's hands and include buttons that the user can press. The motion sensing device may be mounted on the user or on a moving part of the fitness device and contains sensors that capture motion performed by the user. The host device receives input data from the hand controllers and the motion sensing device and generates an activity vector representing the intensity at which the user performs a fitness activity. Based on the activity vector, the host device controls an object in an application, such as a character, based on the input data.

An artificial intelligent agent can act as a player in a video game, such as a racing video game. The game can be completely external to the agent and can run in real time. In this way, the training system is much more like a real world system. The consoles on which the game runs for training the agent are provided in a cloud computing environment. The agents and the trainers can run on other computing devices in the cloud, where the system can choose the trainers and agent compute based on proximity to console, for example. Users can choose the game they want to run and submit code which can be built and deployed to the cloud system. A resource management service can monitor game console resources between human users and research usage and identify experiments for suspension to ensure enough game consoles for human users.

An automated video game testing framework and method includes communicatively coupling an application programming interface (API) to an agent in a video game, where the video game includes a plurality of in-game objects that are native to the video game. The agent is managed as an in-game object of the video game. A test script is executed to control the agent, via the API, to induce gameplay and interrogate one or more target objects selected from the plurality of in-game objects native to the video game. Video game data indicating a behavior of the one or more target objects during the gameplay is received. Based on the received video game data, performance of the video game is evaluated.

A game modification engine modifies configuration settings affecting game play and the user experience in computer games after initial publication of the game, based on device level and game play data associated with a user or cohort of users and on machine-learned relationships between input data and a use metric for the game. The modification is selected to improve performance of the game as measured by the use metric. The modification may be tailored for a user cohort. The game modification engine may define the cohort automatically based on correlations discovered in the input data relative to a defined use metric.

Methods and systems for cooperative or coached gameplay in virtual environments are disclosed. Memory may store a content control profile regarding a set of control input associated with an action in a virtual environment of a digital content title. A request may be received from a set of one or more users associated with different source devices regarding cooperative gameplay of the digital content title. At least one virtual avatar may be generated for an interactive session of the digital content title in response to the request. A plurality of control inputs may be received from the plurality of different source devices and combined into a combination set of control inputs. Generating the combination set of control input may be based on the content control profile. Virtual actions associated with the virtual avatar may be controlled within the virtual environment in accordance with the combination set of control inputs.

Aspects of the present disclosure relate to a personalized agent service that generates and evolves customized agents that can be instantiated in-game to play with users. Machine learning models are trained to control the agent's interactions with the game environment and the user during gameplay. A user may request that a personalized agent join the user's gameplay session. The user device sends a request for the personalized agent to a game platform. The game platform determines whether the user has a license to execute a second instance of the game. When the user has a license to execute a second instance of the game, the second instance of the game may be executed on the user device. Information received from a personalized agent service is used to instantiate a personalized agent in the second instance of the game.

A game management system generates a multidimensional parametric player model representative of player behavior by one or more players in a computer game. The parametric player model populated is used in the identification of groups or clusters of players, and/or in at least partly automated configuration of custom game content for behavior consistent with the parametric player model. A single parametric player model is defines a single set of parametric values corresponding to multiple predefined gameplay parameters, and can be used to model the behavior of a single respective player or to model the behavior of multiple players (e.g., based on cumulative historical gameplay data for the relevant players), providing a representative player model for those players. The player model is ingested by a content generator configured to generate game content customized to the behavior represented by the player model.

A seat system includes a seat including a sensor configured to acquire information for detection of a motion of an occupant seated on a seat body, a terminal device configured to acquire the information from the sensor, and a server capable of communicating with the terminal device. The terminal device executes a game using the sensor based on the information, and acquires an execution result corresponding to the occupant for a game played by the occupant. The server generates integrated data by integrating execution results of the game acquired from another terminal device, and computes a reference value for execution results of the game based on the integrated data. The terminal device or the server assigns a difficulty level of the game for the terminal device based on the execution result corresponding to the occupant and the reference value. The terminal device reflects the difficulty level on the game.