Editorial

AlphaGo vs The World - Thoughts From Your Humble Curators

Back 10 years ago, no one would believe computer Go program can ever beat humans. Many experts estimate it would take 25-50 years to make Go to compete in 9-dan, not become the world champion.

This is exactly what happened yesterday - AlphaGo defeated Ke Jie, the strongest human Go player according to Go Ratings. Go joins the pantheon of games like Chess, where computers have proven to be better than humans. As with Chess, research funding will move from Go to some other more complex A.I.-vs-human research projects.

The big question is - what's the next big game for A.I. to challenge humans? Our guess is the next target is Starcraft/Warcraft. No doubt it would require another sets of technical breakthrough to defeat such complex game with so many game-states.

But before that, deep learning made some real history today.


Other than coverage of AlphaGo (4 items), we also cover SoftBank and statistics of NIPS. As always, if you like our letter, feel free to subscribe/forward it to your colleagues!

Artificial Intelligence and Deep Learning Weekly

News






Blog Posts


Jobs

Video

Videos of AlphaGo vs Ke Jie First Two Matches

Match 1

Match 2

Artificial Intelligence and Deep Learning Weekly

Member's Question

Question by Nishanth Gandhidossā€Ž: Is the following three is what AI is all about to learn for a Data Scientist? Computer vision Natural language processing Reinforcement learning

A: (By Arthur) On the terms - "reinforcement learning" (RL) is usually used as one sub-branch of machine learning, usually goes parallel with "supervised learning" (SL) and "unsupervised learning" (UL). Briefly, RL usually means that you don't have correct output in your training (unlike SL), what you have is just a reward and the reward could be delayed. That makes RL very different from SL, and usually it has its own class of techniques.

"computer vision" (CV) and "natural language processing" (NLP), on the hand, is more applications for machine learnings. You can use techniques from SL, UL and RL in either one of these fields. So that's why the terms CV, NLP and RL seldom compare with each others.

On data scientist's learning of all these subjects - it depends on the type of jobs. More conventionally (say 5 years ago), being data scientist usually means processing data in table forms (R dataframe, SQL etc). But nowadays due to deep learning, data scientist can also be asked to work on high-dimensional data such as computer vision and NLP (through word vector). So I am not surprised that some job description include CV and NLP. If you have some knowledge about them, you do have an edge.

Artificial Intelligence and Deep Learning Weekly

About Us

This newsletter is published by Waikit Lau and Arthur Chan. We also run Facebook's most active A.I. group with 21,000+ members and host a weekly "office hour" on YouTube.

Share on Twitter | Share on Linkedin

Artificial Intelligence and Deep Learning Weekly