Qizheng Pan

Qizheng Pan

Auto-generated profile, based on publicly available information

Qizheng Pan

Xiamen University of Technology, Xiamen, China

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
21 March 2024 in IEEE Transactions on Emerging Topics in Computational Intelligence
Verified
Journal Article
Closed access
YMYing Ma
Ying Ma
XZXiaoyan Zou
Xiaoyan Zou
QPQizheng Pan
Qizheng Pan
Ming YanMing Yan
Ming Yan
GGuoqi Li
Guoqi Li
In the task of multi-label classification, it is a key challenge to determine the correlation between labels. One solution to this is the Target Embedding Autoencoder (TEA), but most TEA-based frameworks have numerous parameters, large models, and high complexity, which makes it difficult to deal with the problem of large-scale learning. To address this issue, we provide a Target Embedding Autoencoder framework based on Knowledge Distillation (KD-TEA) that compresses a Teacher model with large parameters into a small Student model through knowledge distillation. Specifically, KD-TEA transfers the dark knowledge learned from the Teacher model to the Student model. The dark knowledge can provide effective regularization to alleviate the over-fitting problem in the training process, thereby enhancing the generalization ability of the Student model, and better completing the multi-label task. In order to make the Student model learn the knowledge of the Teacher model directly, we improve the distillation loss: KD-TEA uses MSE loss instead of KL divergence loss to improve the performance of the model in multi-label tasks. Experiments on multiple datasets show that our KD-TEA framework is superior to the most advanced multi-label classification methods in both performance and efficiency.

27 December 2022
Verified
Book Chapter
Open access
PDF file
QPQizheng Pan
Qizheng Pan
JLJianmin Li
Jianmin Li
YMYing Ma
Ying Ma
In Partial Multi-Label learning (PML), each training example is assigned with a candidate label set where only partial labels are correct. Existing PML methods only focus on global label correlation, while they lack the consideration of the local label correlation. To alleviate this issue, a novel framework is proposed to jointly consider the feature manifold structure over the global instances and local instances. Specifically, we firstly explore the global feature manifold and local feature manifold by the affinity information conveyed by feature vectors. Then, a trade-off parameter is introduced to character the relative contribution of the feature manifold structures optimized by different methods. Afterwards, in order to disambiguate the candidate labels, we utilize the joint feature manifold in the label space. Finally, the predicted results are learned by training a linear multi-label classification model. Extensive experiments on six PML datasets demonstrate the effectiveness of our proposed method.