Artificial Intelligence is gently but surely taking over the world. You are constantly trying to keep your hand on the pulse, learn new tricks, and at least somewhat anticipate the upcoming trends in machine learning. 

So how do you keep up with all the news and navigate through this endless stream of AI information? 

 

 

First, you need to know what you want and correctly formulate your request. 
Secondly, you need to know where to look and follow to keep up with all the AI trends. And this is where a small list of valuable resources that we have prepared for you will help you. 

Start with...

 

arXiv.org

 
It’s the most important treasure trove of AI/ML knowledge nowadays. Almost all of the papers that come out in the AI world end up here. And it’s no wonder because it’s free, comfortable, and it has simple but handy options like categorization and sorting. And we just love the resource because you can find simply anything here. 

But the AI industry is evolving fast, and the number of new papers and approaches is off the charts. Sometimes it's impossible to keep up with the information flow because there is an ocean of it. So if you go to this resource, you better do it with a specific request. 

Or visit...

Arxiv-sanity & Arxiv-sanity-lite 

To solve the problem of information overflow, five years ago, Andrej Karpathy, Senior Director at Tesla, created Arxiv-sanity, a resource where only the most relevant and high-quality papers end up. It's a kind of filter of goodness. And it’s a brilliant idea in the ocean of endless information. 
Arxiv-sanity has many handy things that are not on ArXiv (or they are counterintuitive). For example, the Top Recent and Top Hype tabs are very convenient and practical.

Also, a pretty nice bonus is the ability to create your own account, where you can add friends, see what they repost, share information, and much more.
However, Andrew decided not to stop there. Instead, not so long ago, he made a complete rewrite of the system, resulting in Arxiv-sanity-lite. 

There’s also...

 

Papers With Code

 

 
When you enter Papers With Code, you first see a beautiful design. Finally!

It's a maximally fabulous resource with a transparent upvote system, allowing you to identify trends quickly. You can find and compare any benchmarks, which, you know, is very important.  You can also find the implementation of the paper itself here. 

Another cool thing about Papers With Code is that you can search for papers and individual approaches in the Methods tab and improve your knowledge of the building blocks of neural networks.

Every 3-4 weeks, the resource posts the fudge of the most breakthrough technologies via newsletter.

Papers With Code has a dataset search and internal portals, where you can follow the world of ML research and CS, Physics, Mathematics, Statistics, and Astronomy.

You can never go wrong with...

 

Twitter

 
Yes, we all know and love Twitter. It's a massive community of people who share their findings. You can always find something interesting here without too much “search stress.”

Of course, the information here is not always the latest and freshest, but you can relax your brain and turn on the “scroll mode.” It’s very helpful from time to time. You just need to lay back, relax, and surf through papers you find worthy. 

My fave account to follow is definitely: @ak92501

You can also check:

Casual GAN Papers

Here you will find a quality breakdown of the latest and most relevant papers.

A 280-character tweet is not always enough for such a task, so the author posts concise news and announcements here.

But that shouldn't stop you. Be sure to go to the author's site, where he explains complex concepts of the latest papers in simple words.

Rivers Have Wings

Yes, it might seem there's no helpful information flow on this channel.  

But something here inspires us to keep doing ML and feel the enormous scale of its amazing possibilities.

Take a look at the AI artworks the channel's creator posts. They are incredible. Fill yourself with love for the industry you're going to be associated with. 

And, finally...

 

YouTube

 
The following channels are worth mentioning.

Yannic Kilcher channel

Here you will find regular in-depth analyses of the latest papers and news in the ML world.

A standard video on the channel is usually 20 to 60 minutes long. So it's the perfect format for anyone who likes to watch useful videos with food.

Two Minute Papers

"What a time to be alive!"

This channel is about 5-minute video reviews of the newest ML novelties. Yes, sometimes they're late, but always of high quality.

StatQuest with Josh Starmer

“StatQuest breaks down complicated Statistics and Machine Learning methods into small, bite-sized pieces that are easy to understand.”

Everything ML beginner might need for a perfect start. 

CodeEmporium

This channel covers essential ML and Data Science concepts.

You will also find the industry insights here.

The channel will appeal to anyone starting in ML.

 
 

 

Staying on top in the raging world of AI news isn’t easy. One can compare the world of AI Research to swimming in the ocean: If you don't know where to swim, you'll just flounder in the vast expanse of water papers. :)

• Learn to make the proper requests.
• Know your resources with relevant data and humane UX.  
• Be carefully inspired by new approaches that are coming out in all industry verticals.

I hope this article was helpful.  

 
 
Bohdan Pytaichuk (@BetterBohdan) is Deep Learning Researcher, AI Evangelist, and Chief Information Officer at AI House. He Holds a Master's degree in Computer Science. Having developed several ML solutions from scratch, he is focused on transforming Ukraine into one of the world's largest AI ecosystems. This activity includes: promoting AI to talented students of the country; educating them; connecting their knowledge with business demands.