best ai feb 2018
March 2, 2018

The Best of AI: New Articles Published This Month (February 2018)

10 data articles handpicked by the Sicara team, just for you

Welcome to the February edition of our best and favorite articles in AI that were published this month. We are a Paris-based company that does Agile data development. This month, we spotted articles about Reinforcement Learning, Data Viz, Capsule Networks and more. We advise you to have a Python environment ready if you want to follow some tutorials :). Let’s kick off with the comic of the month:

self-driving issues

1 — Introduction to Learning to Trade with Reinforcement Learning

learning to trade

As a starter, why not become a master in trading by using machine learning? I found this article interesting as it introduces you to both trading strategies, and Reinforcement Learning. Confront quickly your own RL agent with the stock market!

Read Introduction to Learning to Trade with Reinforcement Learning — from Denny Britz

2 — I wrote some code that automatically checks visualizations for non-colorblind safe colors. Here’s how it works

Color Map Analysis on USA Data
Color Map Analysis on USA Data

8% of the population suffer from color blindness, and I never really thought about which colors I pick for my visualizations. This article gave me lots of insights on how to analyze and produce color maps that are relevant to color-blind and non color-blind people. A must-read if you want to better understand a part of your audience!

Read I wrote some code that automatically checks visualizations for non-colorblind safe colors. Here’s how it works — from Gregor Aisch

3 — Understanding Capsule Networks — AI’s Alluring New Architecture

capsule networks

Capsule Networks became a hot topic in October 2017. With this article, I finally took the time to better understand this new architecture, and why it performs better than CNN on some tasks. If you’re at ease with neural networks, you should definitely take a look at it!

Read Understanding Capsule Networks — AI’s Alluring New Architecture — from Nick Bourdakos

4 — Deep Reinforcement Learning Doesn’t Work Yet

deep reinforcement learning

Last year has been tremendous for Deep Reinforcement Learning. AlphaGo has become the best player in Go, and then AlphaGo Zero has beaten AlphaGo. So why would we consider it not ready yet? Alex Irpan, a Google software engineer explains to us why, according to him, Deep Reinforcement Learning has to be improved.

Read Deep Reinforcement Learning Doesn’t Work Yet — from Alex Irpan

5 — How to think in graphs: an illustrative introduction to Graph Theory and its applications

graph theory

Graph theory might not be the first notion that pops out of your mind when you talk about data science. This article explains how it can be useful on real-world examples (like AirBnB or Twitter). Think about it when you build your next storage architecture.

Read How to think in graphs: an illustrative introduction to Graph Theory and its applications — from Vardan Grigoryan

6 — Family fun with deepfakes. Or how I got my wife onto the Tonight Show

Deepfakes: Nicolas Cage becomes Yoda
Deepfakes: Nicolas Cage becomes Yoda

This month, people have been using neural networks to produce compelling face swaps. Deepfakes are so striking they even got banned from Reddit because of people using them to create fake porn videos. Beyond the ethical issue, it’s always interesting to know how the tech works. Sven Charleer used it to make his wife appear in a famous TV Show, he explains how he did it!

Read Family fun with deepfakes. Or how I got my wife onto the Tonight Show — from Sven Charleer

7 — Bayes’ Rule Applied

sea

I had the opportunity to study Bayes’ Rule during university, but never had the chance to see it applied on a real-world example. This example gave me a step-by-step approach on how to use it. The result is clear as crystal !

By the way, we posted a blog article on Naive Bayes Classifier recently. Make sure to follow us to receive the next ones!

Read Bayes’ Rule Applied — from William Koehrsen

8 — How NASA Earth Observatory creates stunning maps to tell technical stories visually

VIIRS Day/Night Band
Using VIIRS Day/Night Band, Stevens overlays power outages with hurricane Matthew’s progress (left) and uses the information to map the power outages along the coast (right)

NASA is dealing with a large amount of technical data, and one of their major concern is to build visualizations that appeal to non-scientists. Storybench took an interview of Joshua Stevens, lead Data Visualization and Cartography at NASA Earth Observatory and tells us about the process of creating compelling maps.

Read How NASA Earth Observatory creates stunning maps to tell technical stories visually — from Colin Bergmann

9 — Who’s Tweeting from the Oval Office?

trump tweet
Did Trump type out that tweet? Or was it an aide in Trump clothing?

Have you ever asked yourself if Donald Trump is behind every tweet of his account? Greg Rafferty built a model to determine whether Trump wrote it himself or not. In his article, he explains in details his process, from data analysis to model selection. I found it fascinating!

A Twitter bot is live, see it in action @whosintheoval.

Read Who’s Tweeting from the Oval Office?— from Greg Rafferty

10 — A Beginner’s Guide to Data Engineering — Part II

data engineering

I’m a complete newbie when it comes to data engineering and proper ways to store the data. I found this article by Robert Chang from AirBnB to be the perfect fit. Theoretical explanations with real-world code snippets, all we love!

Start with the first step by reading A Beginner’s Guide to Data Engineering — Part I.

Read A Beginner’s Guide to Data Engineering — Part II — from Robert Chang

Thanks to Louis Nicolle, Alexandre Sapet, Vincent Quagliaro, and Geoffroy de Boissieu.

mac mac

5 Mistakes I Made When Doing Custom Data Visualization With D3.js

Experience feedback from drawing graphs in Javascript with D3.js

weird brain

Fast Custom KNN in Sklearn Using Cython

Let’s dive into how you can implement a fast custom KNN in Scikit-learn.

Spark with Jupyter

Get Started with PySpark and Jupyter Notebook in 3 Minutes

Spark is a fast and powerful framework.