Galtier Thomas

Data Scientist - PhD en Mathématiques Appliquées

[Gabriel Peyré @gabrielpeyre]

Reverse mode automatic differentiation computes the gradient with the same cost as the function itself. https://t.co/5op8P77n34

[François Chollet @fchollet]

What machine learning tools do Kaggle champions use? We ran a survey among teams that ranked in the *top 5* of a competition since 2016.

The first question asked about the *primary* framework they used.

Very happy to see confirmation that winning teams prefer Keras :)

[Braden Hancock @bradenjhancock]

I passed my PhD defense today! ("Weak Supervision from High-Level Abstractions"). Main idea: even though ML models (like computers) run on 1s and 0s (individual labels), we can program them using higher-level interfaces. Thank you to so many people who helped me get to this day!

[Alfredo Canziani @alfcnz]

The followings deserve your attention, trust me. When I saw both I was like “Holy 🐄!”!
Jupiter notebook on neural ODE (and useful links!!): https://t.co/2YIhkvxqsh
Blog post on neural ODE: https://t.co/lRMxkfTiKk

[Delip Rao @ COLM26 @deliprao]

This is why we wrote our #nlproc book using barebones PyTorch with good software engineering practices. We were tired of seeing either high level code or poorly written code. Our emphasis is on depth and practice, so you can go create that “superbestlibrary” for your domain/task.

[Galtier Thomas @GaltierThomas]

https://t.co/rGOHy3FjKL

[Albert Pumarola @AlbertPumarola]

We have just released FaSTGAN!! A 32 fps video object segmentation model based on Spatio-Temporal GANs that does not require fine-tuning :)
https://t.co/FtJpaxZ0Z1

[Debprotim Roy @debprotimroy]

Wrote an end to end setup for @fastdotai on GCP for v1. I had trouble finding resources to get this started as directions are quite scattered across blogs + non trivial steps weren't obvious. This should help beginners get started for free.
@jeremyphoward
https://t.co/0aZtCknHgM

[platform ai @platform_ai]

New State-of-the-art in Logo Detection Using YOLOv3 and Darknet https://t.co/8VLCzxFaWa #mlearning #SOTA

[Geoffrey Hinton @geoffreyhinton]

My Coursera MOOC "Neural Networks for Machine Learning" was prepared in 2012 and is now seriously out of date so I have asked them to discontinue the course. But the lectures are still a good introduction to many of the basic ideas and are available at https://t.co/HBbKhA6o8q

[Machine Learning Mastery @TeachTheMachine]

A Gentle Introduction to LSTM Autoencoders https://t.co/ud2eLeesFy

[Sebastian Ruder @seb_ruder]

My PhD thesis Neural Transfer Learning for Natural Language Processing is now online. It includes a general review of transfer learning in NLP as well as new material that I hope will be useful to some.
https://t.co/PxfVRYuyjx

[tkasasagi 🐻 @tkasasagi]

Left: run the model.
Middle: write the paper.
Right: checking citation.

[Sebastian Ruder @seb_ruder]

10 Exciting Ideas of 2018 in NLP: A collection of 10 ideas that I found exciting and impactful this year—and that we'll likely see more of in the future.
https://t.co/iv29bxYbq4

[Sebastian Ruder @seb_ruder]

My AAAI 2019 Highlights—including dialogue, reproducibility, question answering, the Oxford style debate, invited talks, and a diverse set of research papers
https://t.co/18VMsdgmdi

[Andreas Mueller @amuellerml]

My applied machine learning class starts getting to the more meaty bits, we're at preprocessing now: https://t.co/S8UoyYrcRm I realize my slides are still no-where near where I want them to be. But slowly getting there.

[Sebastian Ruder @seb_ruder]

This is a super cool resource: Papers With Code now includes 950+ ML tasks, 500+ evaluation tables (including SOTA results) and 8500+ papers with code. Probably the largest collection of NLP tasks I've seen including 140+ tasks and 100 datasets.
https://t.co/lTAGE7LGZY

[Sebastian Ruder @seb_ruder]

New NLP Newsletter - Challenges in Few-shot learning; 2019 predictions; JAX; Explainable models; MT reading list; Foundations of ML; AI Index 2018; Karen Sparck Jones; Analysis methods survey; ICLR 2019 rejects via @revue https://t.co/EEpM3CVI75