Super excited to announce MLflow, a new open source Machine Learning platform from Databricks to manage the complete machine learning lifecycle: databricks.com/blog/2018/06/0…

Wes Gray 🇺🇸 William Goetzmann What percentage of researchers at the leading quant funds hold a PhD in economics/finance? Today, they are heavily outnumbered by physicists, mathematicians, computer scientists, engineers, etc. They are better trained to analyze complex data, build their own financial theories.

Bloomberg's report on my testimony before the U.S. House of Representatives (December 6, 2019): youtu.be/eFk1nTFaOYQ

Synced: Amazon’s BERT Optimal Subset: 7.9x Faster & 6.3x Smaller Than BERT. syncedreview.com/2020/10/28/ama…

“A gentle dive into the anatomy of a Convolution layer.” Really nice intro from @apiltamang medium.com/@apiltamang/a-…

Some day deep learning might be a science. For now, it remains a craft, so people are rewarded for honing their skills at the craft. #dlearn

I'm co-organizing a new conference on Systems for Machine Learning starting in February; our first call for papers is up at sysml.cc, so submit your interesting SysML work by Jan 5th!

Here's our full paper on improving palliative care with deep learning. Neural net with 18 hidden layers inputs EHR, estimates chance of mortality in next 3-12 months. Also generates report to explain predictions to clinicians. arxiv.org/abs/1711.06402

ML researchers, reviewers, and press coverage of ML need to get a lot more serious about statistically robustness of results and the effect of hyperparameters. This study shows that many papers over the last year or so were just observing sampling error, not true improvement. x.com/karol_kurach/s…

Thank you charu for telling about doing ML at scale. Paytm Labs #gobigorgohome x.com/charupujari/st…

Washington Post has an insightful article on the rise of chatbot therapy, discussing @drwoebot and a few other chatbots. Amy Ellis Nutt washingtonpost.com/news/to-your-h…

#elvesbehavingbadly

Training data is expensive. Image augmentation helps generate more training examples for "free". Small rotations, vertical flips, & cropping, often do not change image labels. This is a library for generating many augmentations. Great idea. github.com/aleju/imgaug

Video of the public lecture I gave at the Institute for Advanced Studies in Princeton on December 12. This talk... fb.me/5PuxFxBh0

Tutorial on “Deep Probabilistic Programming: TensorFlow Distributions and Edward” w/ Rif Saurous, 2pm #POPL2018 conf.researchr.org/track/POPL-201…

Canada.ai would not have been made possible without the invaluable help of our Community Champions: Amii, OR 0R, Vector Institute, @element_ai, CIFAR, MILA, @Next_Canada and Creative Destruction Lab (CDL).

Stanford TensorFlow class by Chip Huyen and Stanford NLP Group web.stanford.edu/class/cs20si/i…

Interesting papers on estimating traces of functions of large matrices (e.g. Hessians) for ML. Very relevant to neural net optimization and large-scale Bayesian inference. Lots of underappreciated ideas from scientific computing! papers.nips.cc/paper/7212-sca… arxiv.org/abs/1802.03451

Blockchain-based Machine Learning Marketplaces medium.com/@FEhrsam/block… I think fitchain.io by ThisIsFrag has a really interesting approach to this.

Best Paper Award #ICLR18: finds flaw in ADAM proof of convergence, shows failure on specific toy convex problem & suggests fix. Given popularity of ADAM, really underscores the need for solid theory. openreview.net/pdf?id=ryQu7f-…