Recently,anincreasing number ofalgorithms for curriculum generation havebeen proposed, empirically demonstrating that CL is an appropriate tool to improve the sample efficiency of DRL algorithms [9, 10]. However, these algorithms are based on heuristics and concepts that are, as ofnow,theoretically notwell understood, preventing theestablishment ofrigorous improvements. In contrast, we propose to generate the curriculum based on a principled inference view on RL. Our approach generates the curriculum based on two quantities: The value function of the agent and the KL divergence to a target distribution of tasks.

Optical flow estimation is a classical yet challenging task in computer vision.

Playful bear cubs and a swirling superpod of dolphins compete for People's Choice honors. Josef has wanted to photograph lynxes for a long time. He was delighted when the opportunity arose to spend two weeks observing them from a hide at Torre de Juan Abad, Ciudad Real, Spain. It's common for young lynxes to play with their prey before killing it. This one repeatedly threw the rodent high in the air and caught it again.

"Gated Orthogonal Recurrent Units: On Learningto Forget".

One simple deferral rule employs the confidence of the current classifier, e.g., based on the maximum predicted softmax probability.