performance analysis of machine learning systems

[Grigori Fursin]

Dear colleagues,

I just wanted to draw your attention to the FastPath workshop on August 23: https://fastpath2020.github.io/Program .

It will feature talks on performance analysis of machine learning systems.

I was also asked to present the key findings from my 10-year quest to enable reproducible ML&systems R&D including artifact evaluations that I have helped to co-organize at MLSys, ASPLOS, CGO, PPoPP, and a few other conferences.

I will also explain why I have developed the Collective Knowledge framework and how it can help to simplify artifact evaluation, reuse artifacts, and automatically co-design efficient ML/SW/HW stacks in terms of speed, accuracy, and different costs particularly focusing on embedded devices with constrained resources.

I will conclude with my plans for the CK framework, the open cKnowledge.io platform, and reproducibility initiatives.

I hope to see some of you there and looking forward to open discussions,
Grigori

===========================================
Grigori Fursin, PhD
Founder of the Collective Knowledge Project
Director of cKnowledge SAS