This week at AIDL:
- Stanford's research on using deep learning to detect skin cancer is one of the biggest news this week. Using deep learning on the medical field is hot. But medical data are intrinsically harder to collect (a disease could be rare), and harder to publish (, rightly so, because of privacy concern).
- Here is a research on predicting patient's future, aptly called Deep Patient. (Thanks to Joseph Bidias.) The model was based on a 3-layer stack of denoising autoencoder, and the researchers found it to outperform other techniques significantly. The article is published in Nature, so I tend to believe its credibility.
- Phys.org report that Apple is going to join Open.AI on Feb 3,I couldn't find other sources to confirm if this news is real, but Apple has every motivation to show that they are into AI, they are, and they have been.
- On fact-checking: this week I looked at whether Google make software which can make software? Faked news is rampant on Facebook, and we should be very careful to source a particular piece of news on AIDL (or anywhere for the matter).
- We have a lot of discussions about H2O.ai . J.T. Bowlin sent out this nifty link of training MNIST.
- Tensorflow is now 1.0.0-rc2, thanks Subhash for the share.
- Talks by Ruslan Salakhutdinov, brought up by Sergey. It's rather exciting among us, because currently there is no solid unsupervised learning tutorial/lecture/seminar series yet
- I updated my Top-5 List since ~3 months. Mostly reflect my belief that Karpathay's class could be a better introduction to students than the Hinton's class, which require much deeper mathematical understanding.
- DeepLearningStars: most starred github projects. Coincidentally, mostly TF-based
- Episode 2 of AID office hour : Greg talks about his work on blockchain and AI.
Must-read: I would read the Stanford's article and Deep Patient's paper in tandem.
If you like this message, subscribe the Grand Janitor Blog's RSS feed. You can also find me (Arthur) at twitter, LinkedIn, Plus, Clarity.fm. Together with Waikit Lau, I maintain the Deep Learning Facebook forum. Also check out my awesome employer: Voci.