GAIT: A Geometric Approach to Information Theory

Gallego-Posada, Jose; Vani, Ankit; Schwarzer, Max; Lacoste-Julien, Simon

doi:10.48550/arxiv.1906.08325

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, inspiring from Maximum State-Visitation Entropy (MSVE), [39] proposes a new reward encouraging an agent to uniformly visit all states of an MDP. Moreover, building on the Geometry Aware Information Theory (GAIT) [32], this measurement of uniformity is defined with respect to a measure of similarity between the states (the geometry of the state space), discouraging the visitation of common states, ie states that are similar to many other states. The authors further propose a simple similarity metric based on a time adjacency principle: states that are close in time are considered similar.…”

Section: Novelty-based Immentioning

confidence: 99%

Autonomous learning in robotics

Wilmot¹

View full text Add to dashboard Cite

Recent advances in artificial neural networks enabled the quick development of new learning algorithms, which, among other things, pave the way to novel robotic applications. Traditionally, robots are programmed by human experts so as to accomplish pre-defined tasks. Such robots must operate in a controlled environment to guarantee repeatability, are designed to solve one unique task and require costly hours of development. In developmental robotics, researchers try to artificially imitate the way living beings acquire their behavior by learning. Learning algorithms are key to conceive versatile and robust robots that can adapt to their environment and solve multiple tasks efficiently. In particular, Reinforcement Learning (RL) studies the acquisition of skills through teaching via rewards. In this thesis, we will introduce RL and present recent advances in RL applied to robotics. We will review Intrinsically Motivated (IM) learning, a special form of RL, and we will apply in particular the Active Efficient Coding (AEC) principle to the learning of active vision. We also propose an overview of Hierarchical Reinforcement Learning (HRL), an other special form of RL, and apply its principle to a robotic manipulation task.

show abstract

Section: Novelty-based Immentioning

confidence: 99%

Autonomous learning in robotics

Wilmot¹

View full text Add to dashboard Cite

show abstract

GAIT: A Geometric Approach to Information Theory

Cited by 1 publication

References 0 publications

Autonomous learning in robotics

Autonomous learning in robotics

Contact Info

Product

Resources

About