Learning Task-Relevant Representations with Selective Contrast for Reinforcement Learning in a Real-World Application

no code implementations • ICML Workshop URL 2021 • Flemming Brieger, Daniel Alexander Braun, Sascha Lange

With the training of the reinforcement learning agent, we present to our knowledge a first approach of using contrastive learning of state-representations for control in a real-world environment, using only images from one static camera.

Contrastive Learning reinforcement-learning +2

Adaptive Blending Units: Trainable Activation Functions for Deep Neural Networks

no code implementations • 26 Jun 2018 • Leon René Sütfeld, Flemming Brieger, Holger Finger, Sonja Füllhase, Gordon Pipa

Since ABUs learn the shape, as well as the overall scaling of the activation function, we also analyze the effects of adaptive scaling in common activation functions.