Construction d'un cadre de lit

Bon j'avoue, je bosse souvent allongé sur un matelas, avec une petite table de lit pour mon ordinateur. Et Quand je ne suis pas jsute dans mon lit, je suis dans mon bureau, ou j'ai installer un petit matelas de fortune par terre…

Bon ça ne paie pas de mine, mais tout de même: quand vous êtes confortablement installé, vous avez de meilleurs chances d'atteindre le “flow”, et moi c'est ce que je recherche ;-)!

→ Read more...

2019/03/15 11:54

Introduction to Deep Q-Networks

Now finally something that sounds a bit more challenging on my Reinforcement Learning journey: Deep Q-Networks! Even if, when you start analysis the structure of those networks, they are still pretty simple in the end. But still, there are 3 key elements that we should consider here: convolutional layers, experience replay, and Target network usage.

On top of that we will also initroduce the Double DQN and Dueling DQN extensions, that are used to improve the system stability and performances. Let's get started!

→ Read more...

2019/03/10 11:36

Model-based policy networks

Continuing on our reinforcement learning path, we will consider here how to build a “model for our environment” and then use this model to train our policy network, instead of the “actual environment”.

→ Read more...

2019/03/09 12:27

Multi-actions vanilla policy Gradient

As a small extension to the previous policy gradient implementation we discussed, we are now going to study how to support multiple actions (ie. num_actions > 2) in the policy network.

→ Read more...

2019/03/09 08:54

Full policy Gradient agent for Reinforcement Learning

This time we are going to handle the creation of a full policy gradient algorithm implementation training on the OpenAI CartPole environment. As opposed to the previous simple policy gradient implementation, this time we will need to handle the previous states to decide what actions to take, and the training network will become sligthly more complex.

→ Read more...

2019/03/08 08:58

Older entries >>