Many interesting things happened since the last Nuit Blanche in Review (September 2016) . First, we had two Paris Machine Learning meetups of very different flavors. The first one took place at vente-privée, a private company that is scaling up its data science team to meet their own amazing scaling issues. The CTO showed us around and most of us were excited at what we saw. The second meetup was a litte different than usual as it was a roundtable around math and Data Science. Unfortunately, one of our main speaker could not stay, but the discussion got a lot of questions and interest. especially around explainability of models and education. We had to stop the debate after a hour and half not because there were no questions left but because the auditorium was booked by a band/choir. As a result, the networking event after the meetup had the musical backdrop of the Star Wars movies while we were all talking about the dark side of algorithms. You could not make that up.
Of note this month on Nuit Blanche, we had a slew of job propositions, four implementations, three theses, a few in-depth articles, and some new hardware description.
Of note this month on Nuit Blanche, we had a slew of job propositions, four implementations, three theses, a few in-depth articles, and some new hardware description.
On LightOn's front, on a regional side of things, we pitched at HTC, and are finalists in the Grand Prix de L'Innovation de Paris. More importantly, we are begining to see an interest in getting to solve the big problems related computations and ML/AI not necessarily connected to our technology. It's a good thing (TM), as it shows a wide interest that is seldom publicized. At the French level, the BioComp GDR had its second colloqium in Lyon. At the European level, David Moloney is organizing a workshop in Dublin next week on Deep Learning and Embedded electronics. In the US, the IEEE has been setting up their Rebooting Computing Initiative ( twitter account is here) while Wired is slowly paying attention to the trend on How AI Is Shaking Up the Chip Market an item that has not escaped specialists in the past few months, well except maybe the good folks at NIPS. Much like the IEEE, it's a just a question of time before this issue becomes more and more important in that comunity as this exchange on Twitter shows:
@NandoDF @susanthesquark @ekoermann which is why I think machine learning papers should include training cost *and* kWh— Nicholas Dronen (@ndronen) 1 octobre 2016
Anyway, we are glad that the future "competitor' questions by VCs will not revolved around Quantum computing only, even though we always have a feel that we're the Barbarians
Enjoy this review !
Sunday Morning Insight
Implementations
- MPI-FAUN: An MPI-Based Framework for Alternating-Updating Nonnegative Matrix Factorization - implementation -
- Input Convex Neural Networks - implementation -
- HyperNetworks - implementation -
- Gaussian graphical models with skggm - implementation -
In-depth
- The Famine of Forte: Few Search Problems Greatly Favor Your Algorithm
- Approximate Sparse Linear Regression
- Universal microbial diagnostics using random DNA probes
- A Greedy Blind Calibration Method for Compressed Sensing with Unknown Sensor Gains
- Practical Learning of Deep Gaussian Processes via Random Fourier Features
- Sketching Meets Random Projection in the Dual: A Provable Recovery Algorithm for Big and High-dimensional Data
- Fastfood Dictionary Learning for Periocular-Based Full Face Hallucination
- Hybrid computing using a neural network with dynamic external memory
Course/Overviews
Thesis
- Thesis: Faster algorithms for convex and combinatorial optimization by Yin Tat Lee.
- Thesis: Spectral Inference Methods on Sparse Graphs: Theory and Applications by Alaa Saade
- Thesis: Fast Randomized Algorithms for Convex Optimization and Statistical Estimation by Mert Pilanci
Paris ML Meetup
- Paris Machine Learning, Hors série #3: Mathématiques et Data Science
- Sunday Morning Insight: Machine Learning in Paris this past week
- Paris Machine Learning Meetup #2 Season 4: Emotional AI, Regaind, Health Knowledge....
Hardware
- Lensless Imaging with Compressive Ultrafast Sensing
- Random Projections for Scaling Machine Learning in Hardware
Job
- Job: PhD Studentships, TU Delft / Online Optimization with Costly and Noisy Measurements using Random Fourier Expansions
- Jobs: Two Postdocs, Statistical Mechanics of Learning, Institute of Theoretical Physics (IPhT), CEA Saclay, France
- Job: Postdoc at Ecole Normale Supérieure (ENS Paris), France
- Job: PhD Studentships, TU Delft / Online Optimization with Costly and Noisy Measurements using Random Fourier Expansions
- CSjob redux: Postdoc, Signal processing & inverse problems for the future SKA, Nice, France
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.
No comments:
Post a Comment