A Configurable Library for Generating and Manipulating Maze Datasets

TL;DR

This paper introduces maze-dataset, a versatile library enabling systematic generation, processing, and visualization of maze datasets to study machine learning models' responses to distributional shifts.

Contribution

The paper presents a configurable library that allows detailed control over maze dataset creation, supporting multiple formats and visualization tools for research on out-of-distribution model behavior.

Findings

Enables systematic study of model responses to distributional shifts.

Supports multiple output formats for diverse model architectures.

Provides tools for visualization and conversion of maze datasets.

Abstract

Understanding how machine learning models respond to distributional shifts is a key research challenge. Mazes serve as an excellent testbed due to varied generation algorithms offering a nuanced platform to simulate both subtle and pronounced distributional shifts. To enable systematic investigations of model behavior on out-of-distribution data, we present $\texttt{maze-dataset}$, a comprehensive library for generating, processing, and visualizing datasets consisting of maze-solving tasks. With this library, researchers can easily create datasets, having extensive control over the generation algorithm used, the parameters fed to the algorithm of choice, and the filters that generated mazes must satisfy. Furthermore, it supports multiple output formats, including rasterized and text-based, catering to convolutional neural networks and autoregressive transformer models. These formats, along with tools for visualizing and converting between them, ensure versatility and adaptability in research applications.