April 2017 was pretty eventful. At LightOn, we just came out with our second newsletter. We attended ICLR in Toulon, and much like Libby , we noticed the millenia CPU paper as well as the current thought processes going on in finding rules of thumbs for new architectures on the algorithm side (see Understanding section below) and on the hardware side (see MLHardware section). We also noted that some of these insights might be automated (see Architecture section). Kudos to the ICLR organizers for streamlining and eventually putting the videos of the event in real time. Other highlights included what is going on in sequencing (see Sunday Morning Insight), a thesis and some jobs. Enjoy !
Sunday Morning Insight:
Implementation (implementation tag)
Architectures (MetaLearning tag)
ML Hardware (MLHardware tag)
- In-Datacenter Performance Analysis of a Tensor Processing Unit (TM)
- MLHardware: Understanding and Optimizing Asynchronous Low-Precision Stochastic Gradient Descent / WRPN: Training and Inference using Wide Reduced-Precision Networks
- It's all about convolutions: YodaNN, TrueNorth computing, FPGAs and a comparison between HOG and CNNs on Hardware
CSHardware (CSHardware tag)
Thesis (Thesis tag)
Highly Technical Reference page:
Jobs: (CSjobs tag)
Credit photo: NASA, JPL-Caltech, SSI, enhanced by Sophia Nasr
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.