Social credit comes up, as the only thing remotely touching the giant undiscussed topic, and the only concrete thing I heard so far is that the narrative in the West is "unfair". Moving on, I guess. #AMLD2019
-
Show this thread
-
I'll just put the HRW report about the use of big data methods in China to identify people to extrajudicially place in indefinite detention in camps here. These are potent technologies—and the research community has to keep all this in mind!
#AMLD2019 https://www.hrw.org/news/2018/02/26/china-big-data-fuels-crackdown-minority-region …pic.twitter.com/lUmVAfyTzJ
3 replies 43 retweets 73 likesShow this thread -
First pic, regular TensorFlow. Second pic, tf-encrypted: a python lib for privacy-preserving ML. This is the track that needs *all* the attention. ML community can’t control that its output will be used for good (it often won’t be) but it can develop tools that do good.
#AMDL2019pic.twitter.com/SKu0G54pdq
3 replies 14 retweets 57 likesShow this thread -
Now, a talk on federated machine learning. Love this track. *This* is stuff the “AI for good” people should work on: first, limit the massive harms on your own tools. The challenges of ML are bias, surveillance, potency at scale, interpretability—often in combination.
#AMLD2019pic.twitter.com/SciUOlQ76H
1 reply 34 retweets 77 likesShow this thread -
Because I keep having a typo in the hashtag, reposting. Here's a comparison between regular TensorFlow code and tf-encrypted—with privacy-preserving python lib addition. I'm constantly surprised how relatively small these research programs are given their importance.
#AMLD2019pic.twitter.com/BTFkaC4Uff
2 replies 15 retweets 34 likesShow this thread -
I love love love the idea of musical interludes at conferences. Plain awesome.
#AMLD2019pic.twitter.com/xrRrAPAZKC
3 replies 14 retweets 92 likesShow this thread -
Great talk on intelligence and machines by
@Kasparov63 but I disagreed with a big chunk of it though I share his optimism but I arrive at it from very different path! It’s actually fun to disagree with a fellow optimist. Will add a few points to my talk tomorrow.
#AMLD2019pic.twitter.com/nXXvxiOTZ0
2 replies 6 retweets 56 likesShow this thread -
Cool presentation by
@NicolasPapernot on machine learning security based on Saltzer and Schroeder principles. The security/privacy track at#AMLD2019 has been robust and the rooms are packed. (As I keep tweeting, this is super important for the research community to prioritize).pic.twitter.com/KfEFrwmnsO
1 reply 8 retweets 27 likesShow this thread -
Almost talk time! Excited and honored to address the research community.

#AMLD2019pic.twitter.com/z4EALna0rr
3 replies 1 retweet 33 likesShow this thread -
Talk done! This slide is something I will write long-form. Bias in ML is a real issue if it reflects existing biases or choice of optimization targets, real solutions are not confined to ML. There are other, urgent research within-ML questions!
#AMLD2019 (pic by@AnnUkhanova)pic.twitter.com/vz8SVMU6lc
3 replies 22 retweets 96 likesShow this thread
zeynep tufekci Retweeted Dylan Richard Muir 💉 💉 💉
Detection *at scale* is a game changer, something that’s (already) restructuring the balance of power between people, corporations and governments. With that, bye #AMLD2019. Thank you for a great experience.https://twitter.com/dylanrmuir/status/1090268562369871875 …
zeynep tufekci added,
-
-
Replying to @zeynep
Please cross post your talk / slides on Linked in and Facebook.
Teşekkür ederim0 replies 0 retweets 0 likesThanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.