MNIST-Gen: A Modular MNIST-Style Dataset Generation Using Hierarchical Semantics, Reinforcement Learning, and Category Theory

TL;DR

MNIST-Gen is an innovative framework that automates the creation of customizable, hierarchical, and semantically meaningful MNIST-style datasets using reinforcement learning, CLIP, and category theory principles.

Contribution

It introduces a modular, adaptive system for generating tailored datasets with minimal manual effort, integrating semantic understanding and hierarchical categorization.

Findings

Achieved 85% automatic categorization accuracy.

Demonstrated 80% time savings over manual dataset creation.

Generated novel datasets: Tree-MNIST and Food-MNIST.

Abstract

Neural networks are often benchmarked using standard datasets such as MNIST, FashionMNIST, or other variants of MNIST, which, while accessible, are limited to generic classes such as digits or clothing items. For researchers working on domain-specific tasks, such as classifying trees, food items, or other real-world objects, these data sets are insufficient and irrelevant. Additionally, creating and publishing a custom dataset can be time consuming, legally constrained, or beyond the scope of individual projects. We present MNIST-Gen, an automated, modular, and adaptive framework for generating MNIST-style image datasets tailored to user-specified categories using hierarchical semantic categorization. The system combines CLIP-based semantic understanding with reinforcement learning and human feedback to achieve intelligent categorization with minimal manual intervention. Our hierarchical approach supports complex category structures with semantic characteristics, enabling fine-grained subcategorization and multiple processing modes: individual review for maximum control, smart batch processing for large datasets, and fast batch processing for rapid creation. Inspired by category theory, MNIST-Gen models each data transformation stage as a composable morphism, enhancing clarity, modularity, and extensibility. As proof of concept, we generate and benchmark two novel datasets-\textit{Tree-MNIST} and \textit{Food-MNIST}-demonstrating MNIST-Gen's utility for producing task-specific evaluation data while achieving 85\% automatic categorization accuracy and 80\% time savings compared to manual approaches.