Author name cluster

Yilin Lyu

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

3 papers

1 author row

NeurIPS Conference 2023 Conference Paper

Overcoming Recency Bias of Normalization Statistics in Continual Learning: Balance and Adaptation

Yilin Lyu
Liyuan Wang
Xingxing Zhang
Zicheng Sun
Hang Su
Jun Zhu
Liping Jing

Continual learning entails learning a sequence of tasks and balancing their knowledge appropriately. With limited access to old training samples, much of the current work in deep neural networks has focused on overcoming catastrophic forgetting of old tasks in gradient-based optimization. However, the normalization layers provide an exception, as they are updated interdependently by the gradient and statistics of currently observed training samples, which require specialized strategies to mitigate recency bias. In this work, we focus on the most popular Batch Normalization (BN) and provide an in-depth theoretical analysis of its sub-optimality in continual learning. Our analysis demonstrates the dilemma between balance and adaptation of BN statistics for incremental tasks, which potentially affects training stability and generalization. Targeting on these particular challenges, we propose Adaptive Balance of BN (AdaB$^2$N), which incorporates appropriately a Bayesian-based strategy to adapt task-wise contributions and a modified momentum to balance BN statistics, corresponding to the training and testing stages. By implementing BN in a continual learning fashion, our approach achieves significant performance gains across a wide range of benchmarks, particularly for the challenging yet realistic online scenarios (e. g. , up to 7. 68\%, 6. 86\% and 4. 26\% on Split CIFAR-10, Split CIFAR-100 and Split Mini-ImageNet, respectively). Our code is available at https: //github. com/lvyilin/AdaB2N.

PDF Details

IJCAI Conference 2023 Conference Paper

Recognizable Information Bottleneck

Yilin Lyu
Xin Liu
Mingyang Song
Xinyue Wang
Yaxin Peng
Tieyong Zeng
Liping Jing

Information Bottlenecks (IBs) learn representations that generalize to unseen data by information compression. However, existing IBs are practically unable to guarantee generalization in real-world scenarios due to the vacuous generalization bound. The recent PAC-Bayes IB uses information complexity instead of information compression to establish a connection with the mutual information generalization bound. However, it requires the computation of expensive second-order curvature, which hinders its practical application. In this paper, we establish the connection between the recognizability of representations and the recent functional conditional mutual information (f-CMI) generalization bound, which is significantly easier to estimate. On this basis we propose a Recognizable Information Bottleneck (RIB) which regularizes the recognizability of representations through a recognizability critic optimized by density ratio matching under the Bregman divergence. Extensive experiments on several commonly used datasets demonstrate the effectiveness of the proposed method in regularizing the model and estimating the generalization gap.

PDF Details DOI

IJCAI Conference 2022 Conference Paper

Learning Target-aware Representation for Visual Tracking via Informative Interactions

Mingzhe Guo
Zhipeng Zhang
Heng Fan
Liping Jing
Yilin Lyu
Bing Li
Weiming Hu

We introduce a novel backbone architecture to improve target-perception ability of feature representation for tracking. Having observed de facto frameworks perform feature matching simply using the backbone outputs for target localization, there is no direct feedback from the matching module to the backbone network, especially the shallow layers. Concretely, only the matching module can directly access the target information, while the representation learning of candidate frame is blind to the reference target. Therefore, the accumulated target-irrelevant interference in shallow stages may degrade the feature quality of deeper layers. In this paper, we approach the problem by conducting multiple branch-wise interactions inside the Siamese-like backbone networks (InBN). The core of InBN is a general interaction modeler (GIM) that injects the target information to different stages of the backbone network, leading to better target-perception of candidate feature representation with negligible computation cost. The proposed GIM module and InBN mechanism are general and applicable to different backbone types including CNN and Transformer for improvements, as evidenced on multiple benchmarks. In particular, the CNN version improves the baseline with 3. 2/6. 9 absolute gains of SUC on LaSOT/TNL2K. The Transformer version obtains SUC of 65. 7/52. 0 on LaSOT/TNL2K, which are on par with recent SOTAs.

PDF Details DOI