Accenture Global Solutions Limited

Dylan James Snow

Robert Dooley

Michael Kuniavsky

Ronak Kumar Bhatia

Implementations are directed to receiving a set of time series image frames within a time period including a plurality of time points; identifying a first entity, wherein the first entity is coupled with a plurality of first positions corresponding to the plurality of time points; identifying a second entity, wherein the second entity is coupled with a plurality of second positions corresponding to the plurality of time points; determining a position difference of the first entity between any two consecutive time points; determining a position difference of the second entity between any two consecutive time points; determining an interaction between the first entity and the second entity based on i) the position difference of the first entity over the time period, and ii) the position difference of the second entity over the time period; determining whether metadata of the interaction satisfies a threshold; and providing feedback on the interaction.

Automated activity detection

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for reinforcement learning in agent training. Agents may be trained using reinforcement learning. The methods, systems, and apparatus include actions of obtaining scene grammars, obtaining a reference video of a reference agent performing a task, generating environments from the scene grammars, determining that behavior of a sample agent shown in a particular environment of the environments matches behavior of the reference agent, storing an indication that the particular environment trains agents to perform the task, determining to train a new agent to perform the task, identifying the particular environment based on the indication, and training the new agent to perform the task in the particular environments identified.

Scene grammar based reinforcement learning in agent training

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for agent environment co-creation. The methods, systems, and apparatus include actions of determining a success rate of an agent in an environment with a first complexity, determining that the success rate satisfies a complexity change criteria, in response to determining that the success rate satisfies the complexity change criteria, determining a second complexity that has a greater complexity than the first complexity, training the agent in the environment with the second complexity, and providing the agent trained in the environment with the second complexity.

Growing community of inventors

Concord, CA, United States of America

Dylan James Snow