Finding Good ML Papers to Read
This will be a short post to hopefully get me to write and post more often. I have mostly found it quite a struggle as time goes by to find more relevant papers to read. The reasons are plenty there is less free time for me to read now than as a graduate student and possibly the other bigger reason there are many many more publications now than there used to be in the areas I care about. See nice plot here of number of adversarial examples publications approaching an exponential function over a short period of time.
Sources I use for finding top papers to read or review
- ML Safety Newsletter AI safety newsletter by Dan Hendrycks.
- Papers with Code Papers and code ranked either by github stars or by social media.
- [https://papers.labml.ai/papers/weekly/] Not entirely sure who runs this but it fetches tweets to rank papers by most discussed.
Sources I want to check out
- Alignment newsletter alignment newsletter by Rohin Shah formerly at CHAI and currently at Deepmind as of the time of writing.
- Import AI Worked on by Jack Clark co-founder of Anthropic talks about different aspects of AI not neccessarily all research related.
- Davis Summarizes Papers Written by Davis Blalock is a research scientist at MosaicML who worked on optimizing ML algorithms.
- Deep Learning Weekly Written mostly be Mikhail Franco Planas who studied industrial engineering and works exclusively on this website as a writer/editor.
AI News and Politics
- Last Week in AI A relatively community run newsletter. Most prominent writers are a Andrey Kurenkov Standford student and Daniel Bashir a compiler engineer.
- EuropeanAI newsletter Covers AI policy in Europe mostly.
- ChinAI Newsletter Covers Chinese policy with a slight focus on AI.
Dead Sources I wish would come back
I used to read r/TopOfArxivSanity.
Please reach out if there are any other good sources that I should look into.