Welcome to the Neural Networks Research Group web site.
The group is directed by Prof. Risto Miikkulainen and is located at the Department of Computer Sciences at the University of Texas at Austin.
Best-paper Honorable Mention at ICDL
5/23/2024
The paper on Using context to adapt to sensor drift, by Jamie Warner, Ashwin Devaraj, and Risto won the "Best-Paper Honorable Mention" at the International Conference on Development and Learning (ICDL-2024, Austin, TX). The idea is to model the context as a separate neural network module, and use it to modulate the outputs of the recognition network. This is a version of the context+skill approach originally proposed by Xun Li for opponent modeling in poker, and used by Cem Tutum and Suhaib Abdulquddos in various control tasks. The code for the sensor drift application as well as the control applications is available at our GitHub repository.