Using playstyle patterns to generate virtual representations of game players

Patent metadata

Field	Value
Publication number	US-10946281-B2
Application number	US-201916369458-A
Country	US
Kind code	B2
Filing date	Mar 29, 2019
Priority date	Mar 29, 2019
Publication date	Mar 16, 2021
Grant date	Mar 16, 2021

Abstract

In various embodiments of the present disclosure, playstyle patterns of players are learned and used to generate virtual representations (“bots”) of users. Systems and methods are disclosed that use game session data (e.g., metadata) from a plurality of game sessions of a game to learn playstyle patterns of users, based on user inputs of the user in view of variables presented within the game sessions. The game session data is applied to one or more machine learning models to learn playstyle patterns of the user for the game, and associated with a user profile of the user. Profile data representative of the user profile is then used to control or instantiate bots of the users, or of categories of users, according to the learned playstyle patterns.

First claim

What is claimed is:

1. A method comprising:
receiving first game session data corresponding to a prior game session of a game application, the prior game session having a first user and a second user as participants;
based at least in part on the first game session data corresponding to the prior game session, executing a new instantiation of the prior game session to include the first user as a participant; and
instantiating a bot corresponding to the second user within the new instantiation of the prior game session, the instantiating including:
determining a subset of the first game session data associated with the second user;
comparing one or more playstyle characteristics of the second user determined using the subset of the first game session data to the one or more playstyle characteristics from one or more user profiles of other users;
selecting one or more machine learning models from the one or more user profiles of the other users based at least in part on the comparing; and
controlling the bot within the new instantiation of the prior game session using the one or more machine learning models and second game session data corresponding to the new instantiation of the prior game session.

2. The method of claim 1 , wherein the prior game session is associated with a game mode of the game application, and the one or more playstyle characteristics are associated with the second user when participating in the game mode of the game application.

3. The method of claim 1 , wherein the new instantiation is started from a requested time within the prior game session.

4. The method of claim 1 , wherein the new instantiation is executed with one or more requested changes to a loadout of the first user.

5. The method of claim 1 , wherein at least one of the one or more machine learning models is trained using at least one of reinforcement learning or inverse reinforcement learning.

6. The method of claim 1 , wherein the one or more playstyle characteristics are determined based at least in part on input data representative of inputs to one or more input/output components of one or more client devices associated with the second user during the plurality of game sessions.

7. The method of claim 1 , wherein the controlling the bot includes using outputs of the one or more machine learning models generated based at least in part on the second game session data.

8. The method of claim 1 , wherein the second game session data comprises image data generated during the new instantiation of the prior game session.

9. A method comprising:
storing first game session data corresponding to a game session of a game application participated in by a first user;
receiving, from a client device associated with a second user, a request to play in a new instantiation of the game session of the game application against a bot corresponding to the first user;
determining a subset of the first game session data associated with the first user;
comparing one or more playstyle characteristics determined using the subset of the first game session data to the one or more playstyle characteristics from one or more user profiles of other users;
selecting one or more machine learning models from the one or more user profiles of the other users based at least in part on the comparing, the one or more machine learning models trained to compute outputs for controlling the bot according to learned playstyle patterns from the one or more user profiles;
executing, based at least in part on the first game session data, the new instantiation of the game session including the second user as a participant and an instantiation of the bot corresponding to the first user; and
controlling the bot within the new instantiation of the game session based at least in part on output data computed using the one or more machine learning models based at least in part on second game session data corresponding to the new instantiation of the game session.

10. The method of claim 9 , wherein the controlling the bot includes applying the second game session data to the one or more machine learning models or one or more additional machine learning models.

11. The method of claim 9 , wherein third game session data used to train the one or more machine learning models includes image data generated during a plurality of previous game sessions of the game application and user inputs mapped to one or more frames of the image data.

12. The method of claim 9 , wherein the request identifies a time within the game session to start the new instantiation, and the new instantiation of the game session is loaded using a subset of the first game session data of the game session representative of the game session at the time.

13. The method of claim 9 , wherein the selecting the one or more machine learning models from the one or more user profiles of the other users is based at least in part on determining that the first user does not have a corresponding user profile.

14. The method of claim 9 , wherein the request to play in the new instantiation of the game session of the game application includes an adjustment to one more characteristics of the game session for the new instantiation of the game session, the one or more characteristics including one or more of items, inventory, health, or location within a game environment for the second user.

15. A method comprising:
training at least one machine learning model to learn aggregate playstyle patterns of a user category based at least in part on historical game session data representative of game sessions of a game application that are participated in by a plurality of users and identified based at least in part on the users being associated with a user category;
receiving, from a device associated with a first user, a request to play in a new instantiation of previously played game session of the game application participated in by the first user and a second user;
executing, based at least in part on first game session data corresponding to the previously played game session, the new instantiation of the game session including the first user and an instantiation of a bot corresponding to the second user; and
determining a subset of the first game session data associated with the second user;
comparing one or more playstyle characteristics determined using the subset of the first game session data to the one or more playstyle characteristics from one or more user profiles of the plurality of users;
selecting the at least one machine learning model from the one or more user profiles of the plurality of users based at least in part on the comparing; and
controlling the bot within the new instantiation of the game session based at least in part on output data computed using the at least one machine learning model and based at least in part on second game session data corresponding to the new instantiation of the game session.

16. The method of claim 15 , further comprising determining the historical game session data associated with the plurality of users, wherein the determining includes, for an individual game session of the game sessions that includes an associated user associated with the user category and an unassociated user not associated with the user category, identifying the historical game session data as a subset of a larger set of game session data for the single game session associated with the associated user.

17. The method of claim 15 , further comprising generating the historical game session data associated with the plurality of users based at least in part on filte

Assignees

• Nvidia Corp

Inventors

• Fear Andrew
• Burke Brian
• Prasad Pillulta Venkata Naga Hanumath
• Lalwani Abhishek

Classifications

• G06N3/045

  Combinations of networks · CPC title

• G06N3/08

  Learning methods · CPC title

• G06N3/006

  based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO] · CPC title

• A63F13/798

  for assessing skills or for ranking players, e.g. for generating a hall of fame · CPC title

• A63F13/795

  for finding other players; for building a team; for providing a buddy list · CPC title