floodsung/deep-learning-papers-reading-roadmap

This repository is a curated reading roadmap of academic papers for anyone wanting to learn deep learning, the branch of machine learning behind modern AI systems like image recognition, speech recognition, and large language models. The problem it solves is that deep learning research spans hundreds of foundational papers published over decades, and a newcomer has no clear guidance on where to start, which papers matter most, or what order makes sense for building up understanding progressively. The roadmap works as a structured list organized into thematic sections: deep learning history and basics, core methods like dropout and batch normalization, specific application domains like computer vision and natural language processing, and more advanced topics like generative adversarial networks and reinforcement learning. Each entry includes the paper's title, authors, a link to the PDF, a brief note on why it matters, and a star rating indicating importance. The sections are ordered deliberately, from foundational survey papers and classic architectures like AlexNet and ResNets, to optimization techniques and cutting-edge methods at the frontier. You would use this repository as a study guide if you are a student, researcher, or engineer starting out in deep learning or trying to build a more rigorous theoretical foundation. Rather than searching Google Scholar randomly, you follow the curated path from first principles through to state-of-the-art research in whatever subfield interests you, whether that is computer vision, speech, natural language processing, or reinforcement learning. Despite being listed as a Python repository, this project contains no runnable code. It is a Markdown document and is effectively a reference bibliography. The value is entirely in the curation and the structured learning path it provides.