AutoNN - An End-to-End Modular Neural Network Framework with GPU-Aware Architecture for Educational and Research Applications

This paper introduces AutoNN, a comprehensive, end-to-end neural network framework implemented entirely in Python, designed to bridge the gap between theoretical understanding and practical implementation in deep learning education and research. AutoNN provides a modular, extensible architecture with full GPU awareness, advanced optimization algorithms, and comprehensive visualization tools, all implemented from scratch without external deep learning library dependencies. The framework supports a wide range of neural network architectures, activation functions, loss functions, and optimization techniques, making it an ideal tool for both educational purposes and rapid prototyping in research environments. Through extensive experimentation, we demonstrate AutoNN's effectiveness across binary classification, multiclass classification, and regression tasks, achieving competitive performance while maintaining transparency and educational value.