Implemented Algorithms

The library contains implementations corresponding to the following papers:

[1] Chuanhao Li, Qingyun Wu, and Hongning Wang.
Unifying Clustered and Non-stationary Bandits. AISTATS 2021.

[2] Chuanhao Li, Qingyun Wu, and Hongning Wang.
When and Whom to Collaborate with in a Changing Environment: A Collaborative Dynamic Bandit Solution. SIGIR 2021.

[3] Chuanhao Li and Hongning Wang.
Asynchronous Upper Confidence Bound Algorithms for Federated Linear Bandits. AISTATS 2022.

[4] Chuanhao Li and Hongning Wang.
Communication Efficient Federated Learning for Generalized Linear Bandits. NeurIPS 2022.

---

Repository Structure

The repository is organized as follows:

• Dataset/
  Datasets used in the experiments.

• dataset_utils/
  Helper scripts/utilities for preparing and handling datasets.

• lib/
  Core library code for bandit algorithms and related utilities.

• SimulationClusteredNonstationary.py
  Script for simulations related to clustered and/or non-stationary bandit settings (primarily for paper [1]).

• SimulationDistributed.py
  Script for simulations in distributed or federated settings (primarily for papers [3] and [4]).

• Articles.py and Users.py
  Problem-specific data structures and logic (e.g., article/user representations).

• conf.py
  Configuration file for experiments (e.g., hyperparameters and other settings).

• custom_errors.py, util_functions.py
  Shared utilities and custom exception definitions.