B
Bingyi Kang
Researcher at National University of Singapore
Publications - 39
Citations - 2112
Bingyi Kang is an academic researcher from National University of Singapore. The author has contributed to research in topics: Computer science & Deep learning. The author has an hindex of 14, co-authored 29 publications receiving 960 citations.
Papers
More filters
Proceedings Article
Decoupling Representation and Classifier for Long-Tailed Recognition
Bingyi Kang,Saining Xie,Marcus Rohrbach,Zhicheng Yan,Albert Gordo,Jiashi Feng,Yannis Kalantidis +6 more
TL;DR: It is shown that it is possible to outperform carefully designed losses, sampling strategies, even complex modules with memory, by using a straightforward approach that decouples representation and classification.
Proceedings ArticleDOI
Few-Shot Object Detection via Feature Reweighting
TL;DR: In this article, a few-shot object detector is proposed that can learn to detect novel objects from only a few annotated examples, using a meta feature learner and a reweighting module within a one-stage detection architecture.
Proceedings ArticleDOI
Overcoming Classifier Imbalance for Long-Tail Object Detection With Balanced Group Softmax
TL;DR: Li et al. as mentioned in this paper proposed a balanced group softmax (BAGS) module for balancing the classifiers within the detection frameworks through group-wise training, which implicitly modulates the training process for the head and tail classes and ensures they are both sufficiently trained.
Proceedings Article
Policy Optimization with Demonstrations
TL;DR: It is shown that POfD induces implicit dynamic reward shaping and brings provable benefits for policy improvement, and can be combined with policy gradient methods to produce state-of-the-art results, as demonstrated experimentally on a range of popular benchmark sparse-reward tasks.
Posted Content
Overcoming Classifier Imbalance for Long-tail Object Detection with Balanced Group Softmax
TL;DR: This work provides the first systematic analysis on the underperformance of state-of-the-art models in front of long-tail distribution and proposes a novel balanced group softmax (BAGS) module for balancing the classifiers within the detection frameworks through group-wise training.