Generalization and Exploration via Randomized Value Functions

Description

This repository contains an unofficial implementation of the Generalization and Exploration via Randomized Value Functions and its application to the Chain environment as in :

• Paper : Generalization and Exploration via Randomized Value Functions
• Authors : I. Osband, B. Van Roy and Z. Wen
• Date : 2016

Details

• Environment : Chain environment (Paper, Figure 1) using bsuite
• Features : Random Coherent basis (Paper, Algorithm 6)
• Evaluation method : Randomized Least Squares Value Iteration (Paper, Algorithm 1) using JAX
• Agent : RLSVI with greedy action (Paper, Algorithm 2)

Usage

To run the experiments :

• Option 1 : Open in Colab.
• Option 2 : Run on local machine. First, you need to clone this repository and execute the following commands to install the required packages :

$ cd randomized-value-iteration
$ pip install -r requirements.txt

You can run an experiment using the following command :

$ cd src
$ python chain.py