reinforcement learning

Lifelong learning is a complex topic in machine learning, whether in combination with reinforce- ment learning or other learning methods. There are different approaches for this and one of them are hypernetworks. Hypernetworks have the advantage of having a very large ability to preserve past memories. Recent work attempts to demonstrate the learning ability of multiple tasks with a hypernetwork such as in the combination of reinforcement learning.

This project adopts Reinforcement Learning to solve a variant of the stochastic VRP, where customers are highly variable, demands are stochastic, and vehicles have a duration limit.

The code in this repository is used for the research on multi-agent reinforcement learning for power grids of Erica van der Sar. The results in the paper "Multi-Agent Reinforcement Learning for Power Grid Topology Optimization" (https://doi.org/10.48550/arXiv.2310.02605) are produced using this code base.

Code and model data for article X (Hautaoja, Kujala and Grahn, 2024. Investigating the Situational Dynamics of Visual Information Sampling in Lateral Vehicle Control: Modeling Driver Sensitivity and Adaptation as Computational Rationality. Journal of Safety Research.

The Simple Pokémon Environment is an AI environment complient with gym that allows for training Reinforcement Learning agents in a simplified version of Pokémon Battles.

Source Code for Gabler et al. "Bayesian Optimization with Unknown Constraints in Graphical Skill-Models for Compliant Manipulation Tasks Using an Industrial Robot", Frontiers in Robotics and AI, 2022

Python project for my master's thesis.

An attempt at developing a learning agent to train an enemy mob in a mario like game

Understanding why DeepMind's DQN is not good enough for some Atari 2600 games.