LPIRC: A Facebook Approach to Benchmarking ML Workload

221 views
Download
  • Share

Learn how Facebook is approaching benchmarking ML for AI and mapping a future of enriched online user experience. Fei Sun, a Software Engineer at Facebook, delves into a systematic approach to benchmark ML on server, mobile and embedded platforms.

Fei Sun says, who am I? If I use three words to describe myself, they are: geek, perfectionist, and workaholic. 

Recently, I'm very interested in deep learning and neural networks. Artificial Intelligence (AI) is gaining popularity these years. Some even forecast that the superintelligence will be real in 30 years. Many still have doubts, but I do not. I believe, the future is here!

I believe, the future economy growth will be catalyzed by the following golden triangle:
- Big data. With the ever increasing amount of data produced every day, it is imperative to filter, rank, and classify those data with automation. Big data is the market, the need.
- Deep learning. With the multi-magnitude dimensional objectives on the data, and gazillions of domains and subdomains, it is vital to find a solution that rules them all. Deep learning is the methodology, the algorithm. 
- Embedded system. With that amount of data, it is essential to process and filter the data in a distributed manner close to the data collector. Embedded system is the foundation, the platform.

https://feisun.org/2017/12/24/a-few-predictions-on-artificial-intellige…
https://feisun.org/2017/12/24/some-scribble-of-things/

Learn how Facebook is approaching benchmarking ML for AI and mapping a future of enriched online user experience. Fei Sun, a ...

Advertisment

Advertisment