Welcome to issue 10 of AI Weekly!
One of the main Machine Learning conferences (ICML 2015) ended a few days ago. Teams from all around the world converged to Lille to discuss novel approaches and challenges.
For a quick summary of the main ideas discussed, have a look at this summary of the discussion between a few of the field's stars (Bengio, Lecun, Schmidhuber, Hassabis...) on the future of Deep Learning, and at this presentation by Leon Bottou (Facebook) on the main challenges facing Machine Learning.
You may also want to have a look at the 2 best papers (according to the jury) of the conference:
- Optimal and Adaptive Algorithms for Online Boosting
- A Nearly-Linear Time Framework for Graph-Structured Sparsity
Have a great week!
In the News
What every manager should know about Machine Learning
Useful for those (managers, friends, coworkers) not familiar with machine learning and who do not necessarily have a heavy tech background.
Robot passes Self-Awareness test for first time
Roboticists at the Ransselaer Polytechnic Institute in New York managed to get one of its robots to pass the 'wise-men puzzle' test of self-awareness, showing that it knows when it is speaking
Gorila: Google Reinforcement Learning Architecture
A hot emerging topic is the industrialisation of Machine Learning. How do you move from experimentation to production-ready code based on ML algorithms? David Silver from Google gave a few clues on Google's approach.
Outlier Detection at Netflix
Slow or unhealthy servers can wreck havoc on Netflix's streaming services. In this blog post they detail their approach to detect such servers. Interesting read.
A Neural Network in 11 lines of Python
A bare bones neural network implementation to describe the inner workings of backpropagation.
Reinforcement Learning - slides
Presentation made by David Silver of Google DeepMind on different approaches to Reinforcement Learning.
Software tools & code
Hadoop & Hive cheatsheets
A few nice cheatsheets for those who work with Hadoop (credit: Bhavya Geethika)
7 Python tools all Data Scientists should know how to use
In case you wonder, the 7 tools are: IPython, GraphLab Create, Pandas, PuLP, Matplotlib, Scikit-learn, Spark.
Machine Learning Can Help Predict Violent Conflicts In Africa
According to the United Nations there is a strong need for robust, accurate, and effective early warning systems for conflict prevention. This article explores ML as a way to predict violence in Africa.
If I were a robot
Robots and AI will become more and more integrated into our daily lives. This post addresses important questions related to this change: how do you teach a robot human behavior? And how do we provide them with social skills?
This newsletter is a weekly collection of AI news and resources. If you find it worthwhile, please forward to your friends and colleagues, or share on your favorite network!
Share on Twitter · Share on Linkedin · Share on Google+
Suggestions or comments are more than welcome, just reply to this email. Thanks!