Deep Learning
Phase diagram of early training dynamics in deep networks: effect of the learning rate, depth, and width
Notably, we discover the opening up of a "sharpness reduction" phase,
Unsupervised Behavior Extraction via Random Intent Priors
Reward-free data is abundant and contains rich prior knowledge of human behaviors, but it is not well exploited by offline reinforcement learning (RL) algorithms. In this paper, we propose UBER, an unsupervised approach to extract useful behaviors from offline reward-free datasets via diversified rewards.
Vulnerabilities in Video Quality Assessment Models: The Challenge of Adversarial Attacks Ao-Xiang Zhang Y u Ran Weixuan T ang Y uan-Gen Wang
No-Reference Video Quality Assessment (NR-VQA) plays an essential role in improving the viewing experience of end-users. Driven by deep learning, recent NR-VQA models based on Convolutional Neural Networks (CNNs) and Transformers have achieved outstanding performance.
