Search Results for author: Lingyu Si

Found 5 papers, 2 papers with code

Disentangle and Remerge: Interventional Knowledge Distillation for Few-Shot Object Detection from A Conditional Causal Perspective

no code implementations26 Aug 2022 Jiangmeng Li, Yanan Zhang, Wenwen Qiang, Lingyu Si, Chengbo Jiao, Xiaohui Hu, Changwen Zheng, Fuchun Sun

To understand the reasons behind this phenomenon, we revisit the learning paradigm of knowledge distillation on the few-shot object detection task from the causal theoretic standpoint, and accordingly, develop a Structural Causal Model.

Few-Shot Learning Few-Shot Object Detection +2

Robust Causal Graph Representation Learning against Confounding Effects

no code implementations18 Aug 2022 Hang Gao, Jiangmeng Li, Wenwen Qiang, Lingyu Si, Bing Xu, Changwen Zheng, Fuchun Sun

This observation reveals that there exist confounders in graphs, which may interfere with the model learning semantic information, and current graph representation learning methods have not eliminated their influence.

Graph Representation Learning

Efficient U-Transformer with Boundary-Aware Loss for Action Segmentation

no code implementations26 May 2022 Dazhao Du, Bing Su, Yu Li, Zhongang Qi, Lingyu Si, Ying Shan

Most state-of-the-art methods focus on designing temporal convolution-based models, but the limitations on modeling long-term temporal dependencies and inflexibility of temporal convolutions limit the potential of these models.

Action Classification Action Segmentation +1

Bootstrapping Informative Graph Augmentation via A Meta Learning Approach

1 code implementation11 Jan 2022 Hang Gao, Jiangmeng Li, Wenwen Qiang, Lingyu Si, Fuchun Sun, Changwen Zheng

To this end, we propose a novel approach to learning a graph augmenter that can generate an augmentation with uniformity and informativeness.

Contrastive Learning Informativeness +2

SimViT: Exploring a Simple Vision Transformer with sliding windows

1 code implementation24 Dec 2021 Gang Li, Di Xu, Xing Cheng, Lingyu Si, Changwen Zheng

Although vision Transformers have achieved excellent performance as backbone models in many vision tasks, most of them intend to capture global relations of all tokens in an image or a window, which disrupts the inherent spatial and local correlations between patches in 2D structure.

Cannot find the paper you are looking for? You can Submit a new open access paper.