NeuralUCB

This repository contains our pytorch implementation of NeuralUCB in the paper Neural Contextual Bandits with UCB-based Exploration (accepted by ICML 2020).

Prerequisites:

• Pytorch and CUDA
• future==0.18.2
• joblib==0.15.1
• numpy==1.18.1
• pkg-resources==0.0.0
• scikit-learn==0.22.1
• scipy==1.4.1
• torch==1.5.0

Usage:

Use python to run train.py for experiments.

Command Line Arguments:

• --size: bandit algorithm time horizon
• --dataset: datasets
• --shuffle: to shuffle the dataset or not
• --seed: random seed for shuffle
• --nu: nu for control variance
• --lambda: lambda for regularization
• --hidden: network hidden size

Usage Examples:

• Run experiments on mnist:

  -  python3 train.py --nu 0.00001 --lamdba 0.00001 --dataset mnist