Awesome Deep Learning Papers

URL: https://github.com/terryum/awesome-deep-learning-papers#readme
Category: Themed Directories
Tags: deep-learning, research, awesome-lists

Description

Awesome Deep Learning Papers is a curated directory of the most-cited and influential deep learning research papers, primarily from 2012–2016. It focuses on seminal works that are considered must-reads for gaining an overview of key developments in deep learning. The list is no longer actively maintained due to the rapid growth of the field.

Features

• Curated top-100 list (2012–2016):

  • Focus on the top 100 most-cited and influential deep learning papers published between 2012 and 2016.
  • Emphasis on classic, seminal works rather than a comprehensive or exhaustive catalog.

• Citation-based inclusion criteria:

  • Papers are evaluated using citation thresholds and impact:
    • < 6 months old: categorized as New Papers (selected by discussion).
    • 2016: ≥ 60 citations or included in a "More Papers from 2016" section.
    • 2015: ≥ 200 citations.
    • 2014: ≥ 400 citations.
    • 2013: ≥ 600 citations.
    • 2012: ≥ 800 citations.
    • Pre-2012: considered Old Papers (selected by discussion).

• Focused on general, seminal contributions:

  • Prioritizes papers whose ideas can be applied across multiple research domains.
  • De-emphasizes narrow application papers, even if highly cited.
  • Selection depends on impact, applicability to other research, and coverage of less-populated research areas.

• Structured sections for paper organization:

  • Top 100 Papers: main curated list.
  • More than Top 100: for important works that did not fit into the top-100 limit.
  • New Papers: recent works (last ~6 months at the time of listing).
  • Old Papers: influential pre-2012 works.
  • More Papers from 2016: an overflow section for noteworthy 2016 papers.

• Dynamic curation rules:

  • The list is strictly capped at 100 main papers.
  • Adding a new paper requires removing another (often from the "More Papers from 2016" section).
  • Removal of papers is treated as a meaningful contribution to maintain quality and focus.

• Context within the ecosystem of deep learning lists:

  • References and complements other themed lists such as:
    • Deep Vision (awesome deep vision papers).
    • Awesome Recurrent Neural Networks.
    • Deep Learning Papers Reading Roadmap (a broader reading sequence for beginners and researchers).
  • Positions itself as a concise, non-overwhelming starting point for understanding core deep learning literature.

• Community contributions (GitHub-based):

  • Open to contributions via pull requests for:
    • Missing or newly important papers.
    • Corrections and updates.
    • Suggestions of key researchers.
  • A dedicated contributing guide is provided in the repository.

• Automation scripts and resources:

  • Script to download all top-100 papers (fetch_papers.py).
  • Script to collect all authors’ names from the top-100 papers.

• Maintenance status:

  • Marked as no longer maintained due to the overwhelming volume of deep learning papers being published since 2017.

Pricing

Not applicable. This is a free, open GitHub-based directory of research papers.