ML news - real-world examples

For inspiration, this thread is to highlight examples of machine learning progress in the real world, both practical and demonstration. Although we might not individually be able to replicate the million dollar projects of Google, Tesla & OpenAI, perhaps there can be discussion on replicating smaller projects, or scaled down versions of the big projects.

As a starter, here is a channel of prolific examples. Perhaps a particular one appeals to you?

So! Where do you get your fix of ML news?

Tip… try watching videos at 2x speed - you learn twice as much. Work your way up… our personal neural net adapts fairly fast. I use this extension in Chrome… (Disclaimer: Most other people will think you are weird.)

5 Likes

PyTorch at Tesla - Andrej Karpathy, Tesla (11min, Nov 2019)

  • Wow! Full Self Driving uses Pytorch!
  • Resnet50 “like” - hey that sounds familiar
  • Hydra sounds a bit like ensembles we learnt about in Lesson 5.

The human inside Tesla’s FSD BETA (17min, Nov 2020)

  • One of my favourite all time FSD clips

Tesla AI Day (2.5hrs, Aug 2021)

1 Like

Twitter! :slight_smile:

8 Likes

As Jeremy mentioned, Twitter.

I also love TheSequence newsletter. It summarizes a lot of stuff going on.

2 Likes

Two Minute Papers is the best. Sometimes I check out the batch.

4 Likes

I believe Andrew Ng also publishes a newsletter , I forget the name though.

2 Likes

Yes, exactly, that’s what @radikubwa mentioned above. It is called The Batch. (How else?)

2 Likes

Also, I quite often see references leading to Mark Tech Post. Here is how they describe themselves:

Marktechpost is a California-based AI News Platform providing 
easy-to-consume, byte size updates in machine learning, deep learning, 
and data science research.

One more thing I tried recently is Alpha Signal newsletter. They rank papers using some ML algorithm and send you a summary. (Note that it has both free and paid versions.)

Though, as was mentioned, Twitter seems to be the greatest source of information. Just need to build a proper feed. I have only a few subscriptions and still getting plenty of interesting links, projects, papers, etc.

2 Likes