Implementation and visualization (some demos) of search and optimization algorithms.

Model-based reinforcement learning using CEM, MPC and PETS

Cross Entropy Method (CEM) implemented under Pytorch, supporting batch dimension and receding horizon style optimization.

CE-ABC is a code to simulate the epidemic outbreaks with mechanistic models through a cross-entropy approximate Bayesian framework.

Solving Tetris using Cross-Entropy Method

Simulation experiments for optimizing objective function with Differential Evolution, Evolution Strategies and Cross Entropy Method (2 versions)

Reinforcement Learning Notebooks

CEopt is a Matlab routine for non-convex optimization using the Cross-Entropy method and augmented Lagrangian formulation.

Simple implementation and comparison of three reinforcement learning models.

Model-Based RL Multi-Tasking with ReLAx

Train a Cross-Entropy Method in Policy-Based Methods with OpenAI Gtm's MountainCarContinous environment

Two dimensional optimisation algorithm using the Cross Entropy Method. Data is iteratively fitted to a Beta Distribution in the algorithm.

Tools for using motion primitives like Dynamic Motion Primitives or Differentiable Linear Dynamic Systems in PyTorch.

SpringpotTune is a Matlab package designed for fitting variable-order springpot models using the Cross-Entropy method.

FraCTune is a Matlab package for tuning fractional-order controllers with the Cross-Entropy method and augmented Lagrangian formulation.

Efficient Model-Based Deep Reinforcement Learning with Predictive Control: Developed a Model-Based RL algorithm using MPC, achieving convergence in 200 episodes (best case) and 1000 episodes on average, outperforming SAC/DQN (10,000+ episodes). Enhanced sample efficiency by 80-90% using learned dynamics and CEM for trajectory optimization.

CROSS-OPT is a Matlab package for optimizing truss structures with the Cross-Entropy method and augmented Lagrangian formulation.

Workshop code for the talk on Introduction to Reinforcement Learning: https://fosterelli.co/file/talk/introduction-to-reinforcement-learning.pdf

Open AI Cartpole environment gradient ascent

Example CEM implementation with ReLAx