Twitter | Search | |
Tiernan Ray
columnist, AI/ML reporter. Tiernan has covered business and finance for over 24 years for Barron’s, Fortune, Bloomberg, The NY Times.
21,033
Tweets
518
Following
10,452
Followers
Tweets
Tiernan Ray 10h
Replying to @yudapearl
They should, a very fine book.
Reply Retweet Like
Tiernan Ray retweeted
ZDNet 12h
Is IBM’s AI demonstration enough for a quantum killer app? by
Reply Retweet Like
Tiernan Ray 14h
Mikael, to test 2,000 drugs initially would be the very definition of brute force. To first make a selection of which 57 drugs to test in combination out of an initial list of 2,000 is a vast reduction in the search space for pre-clinical testing (in vitro).
Reply Retweet Like
Tiernan Ray 14h
Mikael, if you read the article, you will see some ways in which this is not a brute force approach; on the contrary, very selective in the choice of what task to set the machine.
Reply Retweet Like
Tiernan Ray 15h
The importance of Judea Pearl’s counterfactual models for machine learning is highlighted in this Fortune magazine feature on genomics and drug discovery / How A.I. Is Finding New Cures in Old Drugs | Fortune // ⁦
Reply Retweet Like
Tiernan Ray 15h
My article on machine learning transforming pharmacology is live on A story of seasoned genomics researchers changing AI as they change drug discovery // ⁦⁩ ⁦
Reply Retweet Like
Tiernan Ray Mar 17
Replying to @ajratner @ZDNet
Alex, the post has been updated on ZDNet to reflect that Snorkel does generalize to deep networks. Thanks for helping to improve the article.
Reply Retweet Like
Tiernan Ray Mar 17
Replying to @ajratner @ZDNet
That was a sloppy reply. Sorry. DryBell, as opposed to plain Snorkel, is only shown to work on logistic regression. I’d like to see it extended to deep networks. That’s what I meant to say.
Reply Retweet Like
Tiernan Ray Mar 17
Replying to @ajratner @ZDNet
Alex, thanks for pointing it out. My comment in the article was that the DryBell implementation, no plain snorkel, i.e., highly distributed, loosely coupled, so far us only shown on logistic regression, not deep networks. I’d like to see if it can be extended.
Reply Retweet Like
Tiernan Ray Mar 15
Broadcom CEO Hock Tan Proves the Skeptics Wrong With Latest Moves - TheStreet //
Reply Retweet Like
Tiernan Ray Mar 14
Amazing how that trusty convolutional neural network is still conquering new frontiers, including quantum entanglement: Intel offers AI breakthrough in quantum computing | ZDNet ⁦
Reply Retweet Like
Tiernan Ray retweeted
ZDNet Mar 14
Intel offers AI breakthrough in quantum computing by
Reply Retweet Like
Tiernan Ray Mar 14
Replying to @siwanoy
Nicely put, Brian.
Reply Retweet Like
Tiernan Ray Mar 13
Lyft Will Be Relying on One Unorthodox Number to Sell Its IPO - TheStreet //
Reply Retweet Like
Tiernan Ray retweeted
Nelson Wang Mar 13
Lyft faces the tall order of emulating Amazon to succeed; nice analysis from via
Reply Retweet Like
Tiernan Ray Mar 12
Talend One-on-One CEO Interview - Expects Cloud Sales to 'be a very, very strong grower for a number of years' - //
Reply Retweet Like
Tiernan Ray retweeted
ZDNet Mar 12
Please, Facebook, give these chatbots a subtext! by
Reply Retweet Like
Tiernan Ray Mar 11
Google's “TF-Replicator” technology: just a little bit of Python code to build distributed machine learning; training ResNet-50 in under half an hour | ZDNet // ⁦
Reply Retweet Like
Tiernan Ray Mar 11
Amazon Has a Fulfillment Cost Issue It Needs to Resolve - TheStreet //
Reply Retweet Like
Tiernan Ray retweeted
ZDNet Mar 8
Google's distributed computing for dummies trains ResNet-50 under half an hour by
Reply Retweet Like