The virtual-to-real paradigm, ie, training models on virtual data and then applying them to
solve real-world problems, has attracted more and more attention from various domains by …

Developing autonomous driving technologies necessitates addressing safety and cost
concerns. Both academic research and commercial applications of autonomous driving …

The development of machine learning in complex system is hindered by two problems
nowadays. The first problem is the inefficiency of exploration in state and action space …

One of the main challenges of advancing task-oriented learning such as visual task planning
and reinforcement learning is the lack of realistic and standardized environments for training …

Hyperdimensional computing (HDC) is an emerging learning paradigm that computes with
high dimensional binary vectors. There is an active line of research on HDC in the …

Pre-training has become a standard paradigm in many computer vision tasks. However,
most of the methods are generally designed on the RGB image domain. Due to the …

Deep learning based semi-supervised learning (SSL) algorithms have led to promising
results in recent years. However, they tend to introduce multiple tunable hyper-parameters …

One of the core tasks in multi-view learning is to capture relations among views. For
sequential data, the relations not only span across views, but also extend throughout the …

Mainstream 3D representation learning approaches are built upon contrastive or generative
modeling pretext tasks, where great improvements in performance on various downstream …

Transferring knowledge from pre-trained deep models for downstream tasks, particularly
with limited labeled samples, is a fundamental problem in computer vision research. Recent …