State Representation Learning

for

Reinforcement Learning

Antonin Raffin, Natalia Diaz-Rodriguez, Ashley Hill, René Traoré and David Filliat

INRIA Flowers Deep Reinforcement Learning Workshop

5th April 2018 - Paris

Dataset	Ground Truth	Robotic Priors	Autoencoder	VAE	PCA
Baxter + Distractors	0.024	0.079	0.099	N.A.	N.A.
Kuka - static button	0.00248	0.00279	0.00281	0.00281	0.00425

State Representation Learning for Reinforcement Learning Antonin Raffin , Natalia Diaz-Rodriguez, Ashley Hill, René Traoré and David Filliat INRIA Flowers Deep Reinforcement Learning Workshop 5th April 2018 - Paris

State Representation Learning for Reinforcement Learning

By Antonin Raffin

State Representation Learning for Reinforcement Learning

7 years ago
1,687

Antonin Raffin

araffin.github.io
araffin2

State Representation Learning

for

Reinforcement Learning

Outline

State Representation Learning (SRL)

Methods

Robotic Priors

Environments

Learned States (Baxter)

Learned States (Kuka)

Evaluation

1. KNN-MSE

2. Reinforcement Learning

KNN-MSE

RL Algorithms: OpenAI Baselines

RL Setting

Learning Curve

Tips and Tricks

Software Stack

Simulator Fight

VS

Visualization: Vizdom

Workflow: Git

Organization: Trello

Conclusion

Thank You!

State Representation Learning for Reinforcement Learning

State Representation Learning for Reinforcement Learning

Antonin Raffin

State Representation Learning

for

Reinforcement Learning

State Representation Learning for Reinforcement Learning

More from Antonin Raffin