Welcome to the latest 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 Deep learning, Reinforcement learning, Graph generation, and a debate around the AI regulation. We sure provided something for your taste! Let’s kick off with the comic of the month:
Why does deep learning work so well? “Information bottleneck” could be the key to finally unlock this mystery. The main idea is that deep learning works as a bottleneck that get rid of all noisy information to retain only highly relevant features.
This is the first part of an exciting tutorial about reinforcement learning with Starcraft II environment. In this one, you will learn how to set up your environment and train your first model.
Are you looking for a quick and easy way to analyze social posts? This tutorial shows you, through Donald Trump’s tweets, how to extract such data, do some stats and visualization. Finally you will conduct sentiment analysis. All in 30 minutes.
Data science is not just about taking a nice dataset, plugging an ad hoc learner from whatever python library and training a model. Real-world problems are much more challenging. This article presents all the fallacies you won’t face at university or Kaggle competitions but are part of your journey as a real data scientist.
With this tutorial you will learn to generate amazing graphics only with matplotlib and pandas. You will begin with a banal and boring matplotlib graph that everybody is used to. You will end up with an outstanding result.
Object detection is probably the most challenging branch of Computer vision. It aims at detecting and classifying objects in an image. With the breakthroughs of deep learning, it has evolved a lot. If you’re interesting in the history of object detection and state of the art algorithms (YOLO, Faster R-CNN, F-RCN…), this article is for you!
You certainly are aware of the Three Rules of Robotics formulated by Isaac Asimov in 1942. Following recent debates on the dangers of A.I. and the need for its regulation, Oren Etzioni attempts to propose The Three Rules of A.I.
Bokeh vs Dash — Which is the Best Dashboard Framework for Python?
This article compares Bokeh and Dash (by Plotly), two Python alternatives for the Shiny framework for R, using the same example.
Few-Shot Image Classification with Meta-Learning
Here is how you can teach your model to learn quickly from a few examples.
How to Track your Users over Several Domains?
Track users over different domains is a recurrent issue while developing a substantial web solution.