Twitter | Search | |
Gerard Sanroma
computer vision and machine learning
1,182
Tweets
198
Following
172
Followers
Tweets
Gerard Sanroma retweeted
Andrew Gordon Wilson Feb 20
Our new paper "Bayesian Deep Learning and a Probabilistic Perspective of Generalization": . Includes (1) benefits of BMA; (2) BMA <-> Deep Ensembles; (3) new methods; (4) BNN priors; (5) generalization in DL; (6) tempering in BDL. With . 1/19
Reply Retweet Like
Gerard Sanroma retweeted
François Chollet Feb 20
The number one thing to keep in mind about machine learning is that performance is evaluated on samples from one dataset, but the model is used in production on samples that may not necessarily follow the same characteristics...
Reply Retweet Like
Gerard Sanroma retweeted
Sayak Paul Feb 17
Blow your mind: .
Reply Retweet Like
Gerard Sanroma retweeted
DeepAI Feb 16
Reply Retweet Like
Gerard Sanroma retweeted
Stephen Smith Feb 13
New UK Biobank data release: brain imaging data (raw + preprocessed + IDPs) from 40,000 participants. Plus second-scan (repeat imaging data) from 1,500 of those participants. More details at
Reply Retweet Like
Gerard Sanroma retweeted
Microsoft Research Feb 10
Microsoft researchers and engineers release Zero Redundancy Optimizer (ZeRO) and DeepSpeed library, a system able to train 100-billion-parameter deep learning models. Learn about this breakthrough and how it led to Turing Natural Language Generation:
Reply Retweet Like
Gerard Sanroma Feb 9
Replying to @saulfp @VisiData
Cool! Ctrl + Home/End could be used instead to go to beginning/end vertically. If I remember correctly, that‘s how txt editors used to be?
Reply Retweet Like
Gerard Sanroma retweeted
Facebook AI Feb 6
We just released PyTorch3D, a new toolkit for researchers and engineers that’s fast and modular for 3D deep learning research:
Reply Retweet Like
Gerard Sanroma retweeted
Bernhard Schölkopf Feb 5
„Data science and machine learning conferences (NeurIPS, ICML, AISTATS, ICLR, UAI, ...): Allow remote paper & poster presentations at conferences” -
Reply Retweet Like
Gerard Sanroma retweeted
Wouter van Amsterdam Feb 5
Very interesting paper on causality for medical imaging! With causal diagrams we can exactly define pitfalls that hamper external validity and eventual usability in medical imaging. Definition leads to understanding and the possibility of solutions. Must read for ppl in med img.
Reply Retweet Like
Gerard Sanroma retweeted
Junpeng Lao🇨🇭 Feb 4
So in the podcast I told Alex that I wish there are more writings on the design of the TFP MCMC component, as I grow to love how expressive it is. Happy to share what we (me, , and many others) submitted to
Reply Retweet Like
Gerard Sanroma Feb 3
Replying to @roydanroy
Price of Spanish or Italian wine in Britain
Reply Retweet Like
Gerard Sanroma retweeted
Ted McCormick Jan 28
My wife and I never talk to our kids. We ask them to submit all communications anonymously through rigorous and double-blind peer review. My daughter is 5 and has 21 publications in high-impact journals. My daughter can burn holes in furniture using only her eyes
Reply Retweet Like
Gerard Sanroma retweeted
Eric Topol Jan 28
My 1st -guided echocardiography today (fun & by democratizing acquisition it will change the field) What does that mean? You put the transducer on the patient's chest ->the AI tells you how to properly position it and then automatically captures video images when spot on.
Reply Retweet Like
Gerard Sanroma Jan 28
Replying to @saulfp @VisiData
as a side note: it is usually easier to search for columns than rows in case the mistake was the other way round
Reply Retweet Like
Gerard Sanroma Jan 28
Replying to @saulfp @VisiData
I expect 'Home' and 'End' keys to move the cursor horizontally (ie, column-wise) but rather they move it vertically. Need to search back for the lost row. Not a big thing but now you ask :)
Reply Retweet Like
Gerard Sanroma Jan 27
Replying to @dan_s_becker
Looks interesting. What techniques do you use for combining ML and rules?
Reply Retweet Like
Gerard Sanroma Jan 26
Mee too :)
Reply Retweet Like
Gerard Sanroma retweeted
Aurelio Ruiz Jan 24
Interesting survey on machine and statistical learning for longitudinal analysis of neuroimaging data in Alzheimer’s Disease by and Gemma Piella at in & . Methods, data and future directions of the field
Reply Retweet Like
Gerard Sanroma retweeted
NumPy Jan 23
Hello, world! This is the official account for NumPy, the fundamental package for scientific computing with Python. Follow us for news, info and content related to NumPy!
Reply Retweet Like