Where I Find My Deep Learning News
It’s hard to find reliable sources of information when starting out in deep learning. These will save you time, effort, and headaches.
See also: 2022 Update
There are tons of free machine learning resources out there, but the sheer volume makes it difficult to sift through them. Not all of them are reliable or well-written. When you’re starting out in deep learning, a poorly written tutorial can do more harm than good. I sincerely hope that never happens to you, but it’s affected me more than a few times.
These are my favorite resources for deep learning news, which I’ve accumulated over the past 4 years working as a Machine Learning Engineer. I love them, because they save me time and effort. There are simply not enough hours in a week to manually scour the ArXiv for research papers. These sites can help to filter that down, and show you only the high-quality content you need.
I also publish a weekly update on deep learning, which contains research papers, blog posts, and Github repos I liked from the past week. Follow me to get a condensed, curated list of deep learning news each week! You can check out recent editions here:
- Keeping Up with Deep Learning — 3 Dec 2020
- Keeping Up with Deep Learning — 26 Nov 2020
- Keeping Up with Deep Learning — 19 Nov 2020
Papers with Code
[Home Page]
Papers with Code is the single most useful resource for deep learning right now. It aggregates papers from ArXiv, and lists open source implementations from Github for each paper. The Trending Research page sorts projects by the number of Github starts per hour, so you can easily find the most popular, recent papers. The Browse State-of-the-Art tab organizes the papers into different sub-domains, such as computer vision or natural language processing. Each sub-domain contains benchmarks and leaderboards to help identify the best methods available.
ArXiV Sanity Preserver
[Home Page][YouTube]
ArXiv Sanity Preserver is another tremendous time-saver, because it aggregates recent and popular ML papers from the ArXiv. It is created and maintained by Andrej Karpathy (head of AI at Tesla) in his spare time, with the goal of accelerating research in deep learning. My favorite feature from ArXiv Sanity is the Recommend tab. Once you save some papers to your library, it will automatically recommend similar papers that were recently published!
Connected Papers
[Home Page]
Connected Papers helps uncover research papers related to the ones you’re already reading. It visualizes ArXiv papers as a graph, where each connection represents a cross-reference between two papers. Influential papers tend to have more cross-references, which makes them the central part of each graph. You can find related papers by looking at neighbors in the graph, and easy to explore all of the related literature for a particular topic.
When used correctly, Twitter can be a great source of news for any subject area. You just have to follow the right people. Deep learning experts from industry and academia regularly tweet about their recent work, and share notable work from their friends and colleagues. If you’re not sure who to follow, start with just one well-known person in deep learning — maybe Yann LeCun, Turing Award winner and head of AI at Facebook. Look at who they follow, and follow some of the AI folks yourself. You can also check out my following list to get more ideas.
Top Medium Writers
[Deep Learning Tag][Artificial Intelligence Tag]
This might go without saying, but there are many writers on Medium who focus on deep learning. You can find the top stories and writers for the Deep Learning tag on Medium. I personally follow Jonathan Hui, Richmond Alake, and Jingles (Hong Jing), and I can recommend them with confidence.
GitHub
Similar to Twitter, you can follow your favorite programmers and researchers on GitHub. But instead of tweets, you’ll get notifications each time they create a repo or push a new project release. Follow the people whose projects you’ve used, and you’ll quickly build a personalized feed of open source updates.
Friends and Colleagues
Your close friends and colleagues are probably the best source of all, because they understand your needs and interests better than any recommendation algorithm. I have several text threads with friends, where we discuss and recommend papers (among other things). My work Slack is full of links to interesting articles, and we recently started a reading group to work through Yann LeCun’s deep learning course from NYU. to interesting papers and articles. These relationships provide many other benefits that you can’t get from a website.
- Grow your professional relationships.
- Strengthen your personal and professional network.
- Give and receive encouragement for current projects.
- Develop a sense of community and camaraderie.
Conclusion
As someone without a formal education in machine learning, it took some time for me to identify high-quality sources of deep learning news. I hope this helps other people to avoid the same problem. And if you think anything should be added to the list, please let me know!
