Project: Deep Learning Phenomena

Motivation

Deep learning has achieved great success, but understanding how it works is essential for building better and safer AI systems. Deep learning is a highly complex system. Studying phenomena—observable patterns in experiments—is a key way to understand such complexity. This approach mirrors how scientists study nature: through controlled experiments, they discover empirical laws that guide theory, which in turn leads to deeper understanding and useful technology.

Objectives

This project collects important phenomena in deep learning and provides a platform for researchers to share and discuss their findings. By gathering well-documented phenomena with reproducible experiments, we aim to help researchers identify open questions, connect theory with practice, and advance the field together.

Contributions

This project is a collaborative effort of researchers from various institutions. We welcome contributions from researchers and practitioners in the field of deep learning. You can contribute a phenomenon by submitting a Markdown file to the project, or submit an information of a related paper to an existing phenomenon. Please see the Submission Guide for more details. We encourage researchers to share their findings and contribute to the collective understanding of deep learning phenomena.