This Week in Machine Learning: Hope for Parkinson’s Disease, AI Threat, and AI Economics

Posted August 18, 2020

It’s been a while so we’re coming back with another dose of machine learning news. A lot has happened so let’s get right to it!

Here’s a bit of the latest news, discoveries, and interesting articles. Enjoy the read!

Bi-Weekly Roundup: August 3-17

» blog – as always, make sure to visit our blog to find out interesting and in-depth articles on machine learning from the last week. 🙂

» IBM and Michael J. Fox Foundation develop machine learning model for Parkinson’s by Aimee Chanthadavong | August 7

Parkinson’s disease is one of the cruelest for the human brain. Hopefully, there’s ML. Recently, the research arm of Big Blue and The Michael J. Fox Foundation (MJFF) have developed a machine learning disease progression model as part of further research into Parkinson’s disease (PD). 🧠

» 2035’S BIGGEST A.I. THREAT IS ALREADY HERE by Sarah Wells on Invers | August 7

A team of academics, policy experts, and private sector stakeholders warn there is trouble on the horizon. They’ve pinpointed the top 18 artificial intelligence threats we should be worried about in the next 15 years. What are the threats? We’re too scared to say, just check it out yourself… or skip this news 😉

» Taming the Tail: Adventures in Improving AI Economics by Martin Casado and Matt Bornstein | August 13

A concise and practical guide on how to tackle the challenges of AI business (especially the economic ones). Lessons, best practices, and earned secrets. 🤖

» Amazon’s Machine Learning University is making its online courses available to the public by Douglas Gantenbein on Amazon | August 12

Classes previously only available to Amazon employees will now be available to the community. Beginning in 2021, all MLU classes will be available via on-demand video, along with associated coding materials. Make sure to check it out, maybe you’ll find something you like! 👨‍💻

» A.I. can tell if you’re a good surgeon just by scanning your brain by Luke Dormehl on | August 12

Researchers at Rensselaer Polytechnic Institute and the University at Buffalo have developed Brain-NET, a deep learning A.I. tool that can accurately predict a surgeon’s certification scores based on their neuroimaging data. Interesting! 😷

» How Facebook’s Yann LeCun is charting a path to human-level artificial intelligence by Thomas Macaulay on The Next Web | August 14

If you’re a fan of Yann LeCun, Facebook’s chief AI scientist, make sure to read the article in which he told TNW about his work at FAIR, the social network’s research lab.

» KDnuggets™ News of the week with top stories and tweets of the past week, plus opinions, tutorials, events, webinars, meetings, and jobs.

Here’s KDnuggets™ News of the week from two weeks ago

» Old but gold, the reliable Reddit thread on ML for more news on machine learning. There’s always something for everyone – tips, tricks, hacks, and more news.

That’s all folks! I hope you found something of interest in this weekly roundup. Don’t forget to check our blog for more inspiring articles.

👉 Came across an interesting ML article? Or maybe you wrote one yourself and would like to share it with other people? Let us know, we’ll spread the news in our weekly roundup!


ML Experiment Tracking: What It Is, Why It Matters, and How to Implement It

Jakub Czakon | Posted November 26, 2020

Let me share a story that I’ve heard too many times.

”… We were developing an ML model with my team, we ran a lot of experiments and got promising results…

…unfortunately, we couldn’t tell exactly what performed best because we forgot to save some model parameters and dataset versions…

…after a few weeks, we weren’t even sure what we have actually tried and we needed to re-run pretty much everything”

– unfortunate ML researcher.

And the truth is, when you develop ML models you will run a lot of experiments.

Those experiments may:

  • use different models and model hyperparameters
  • use different training or evaluation data, 
  • run different code (including this small change that you wanted to test quickly)
  • run the same code in a different environment (not knowing which PyTorch or Tensorflow version was installed)

And as a result, they can produce completely different evaluation metrics. 

Keeping track of all that information can very quickly become really hard. Especially if you want to organize and compare those experiments and feel confident that you know which setup produced the best result.  

This is where ML experiment tracking comes in. 

Continue reading ->