Twitter | Search | |
Loreto Parisi
MSc Computer engineering. Artificial Intelligence and Machine Learning Father of an amazing girl and a cute boy.
11,394
Tweets
624
Following
861
Followers
Tweets
Loreto Parisi retweeted
Luigi Acerbi 18h
My new work accepted at is finally available! VBMC is a novel machine learning method to perform Bayesian posterior and model inference when the model likelihood is moderately expensive to evaluate. (continued...)
Reply Retweet Like
Loreto Parisi retweeted
topcoder 21h
The $19,000 Mozilla Bugzilla Data Science Challenge has one week left! Learn and register now! #DataSciencehttps://www.topcoder.com/community/data-science/Mozilla-Bugzilla-Data-Science-Match
Reply Retweet Like
Loreto Parisi 20h
“The doesn’t belong to the fainthearted. It belongs to the brave”.
Reply Retweet Like
Loreto Parisi 24h
In the Adversarial Attack is a problem for models, especially when image classification has been deployed to crucial applications. A perturbation imperceptible to humans - Fast Gradient Sign method FGSM can fool the network
Reply Retweet Like
Loreto Parisi Oct 13
Replying to @Reza_Zadeh
It's pretty interesting that this unsupervised approach follows the neuron work, where while training a RNN a sentiment feature was found at the text feature 2388
Reply Retweet Like
Loreto Parisi retweeted
Reza Zadeh Oct 13
Two unsupervised tasks together give a great features for many language tasks: Task 1: Fill in the blank. Task 2: Does sentence X immediately follow sentence A? Transfer learning showing great strides in language, as it does with Computer Vision. Paper:
Reply Retweet Like
Loreto Parisi retweeted
Sebastian Ruder Oct 12
It's amazing how fast is moving these days. We have now reached super-human performance on SWAG, a commonsense task that will only be introduced at in November! We need even more challenging tasks! BERT: SWAG:
Reply Retweet Like
Loreto Parisi Oct 12
will word-level work better than char-level in neural machine translation with seq2seq?
Reply Retweet Like
Loreto Parisi retweeted
hardmaru Oct 11
“Recycle-GAN”: Unsupervised Video Retargeting. Translation from John Oliver to Stephen Colbert, and a synthesized flower follows the blooming process with the input flower. Website has more video demos demonstrating video retargeting for faces and flowers:
Reply Retweet Like
Loreto Parisi retweeted
Rachel Thomas Oct 11
A Review of the Neural History of Natural Language Processing-- great neural NLP overview by
Reply Retweet Like
Loreto Parisi retweeted
TensorFlow Oct 11
In this guest post on the TensorFlow blog, VP Data Science & Analytics at BHGE Arun Subramaniyan explores industrial ML with probabilistic deep learning andTensorFlow Probability. Read the post here ↓
Reply Retweet Like
Loreto Parisi Oct 11
How to correctly configure Tornado application, access, general loggers dictionaries
Reply Retweet Like
Loreto Parisi retweeted
Andrea Benedetti Oct 11
(AI) is accelerating the for every industry, with examples spanning manufacturing, retail, finance, healthcare and many others Download & read the : A Developer's Guide to Building AI Applications
Reply Retweet Like
Loreto Parisi Oct 11
guys your bikes have a problem fix it! "A Distributed Reinforcement Learning Solution With Knowledge Transfer Capability for A Bike Rebalancing Problem"
Reply Retweet Like
Loreto Parisi retweeted
Quanta Magazine Oct 8
A graduate student has solved a fundamental question in quantum computation: How can a non-quantum observer verify the work of a quantum computer?
Reply Retweet Like
Loreto Parisi retweeted
Lin Clark Oct 8
Function calls between JavaScript and WebAssembly are finally fast! 🎉 wasm » JS 🔥 ~750ms → 450ms JS » wasm 🔥 ~5500ms → 450ms monomorphic JS » wasm 💥 ~5250ms → 250ms wasm » built-in 🔥 ~5750ms → 600ms See how we did it →
Reply Retweet Like
Loreto Parisi Oct 8
Replying to @maxpumperla @fchollet
Yep I’m using hyperopt with fasttext but I like the idea to keep this Keras tooling as simpler as possibile.
Reply Retweet Like
Loreto Parisi retweeted
François Chollet Oct 7
Here's a new open-source package for hyperparameter search for Keras models: (no affiliation with the authors, I just thought it was cool)
Reply Retweet Like
Loreto Parisi retweeted
Peter Martigny Aug 14
How to do NLP classification if you don't have much labeled data? See how the team uses the ULMFit paper from to classify Amazon reviews, Great performance with only 100 samples :)
Reply Retweet Like
Loreto Parisi retweeted
Jeremy Howard Aug 14
This is a terrific introduction to NLP transfer learning, showing how to match fasttext's results on 4 million Amazon reviews using just 1000 reviews with ULMFit!
Reply Retweet Like