Confidence correction for trained graph convolutional networks

Yuan, Junqing1; Guo, Huanlei1,2; Zhou, Chenyi3; Ding, Jiajun4,5,6; Kuang, Zhenzhong4; Yu, Zhou4; Liu, Yuan7

2024-12-01

10.1016/j.patcog.2024.110773

PATTERN RECOGNITION   影响因子和分区

0031-3203

1873-5142

Adopting Graph Convolutional Networks (GCNs) for transductive node classification is a hot research direction in artificial intelligence. Vanilla GCNs are primarily under-confident and struggle to clarify the final classification results explicitly due to the lack of supervision. Existing works mainly alleviated this issue by improving annotation deficiency and introducing addition regularization terms. However, these methods need to re-train the model from the beginning, which is computationally expensive for large dataset and model. To deal with this problem, a novel confidence correction mechanism (CCM) for trained GCNs is proposed in this work. Such mechanism aims at calibrating the confidence output of each node in the inference stage by jointly inferring the feature and predicted pseudo label. Specifically, in the inference stage, it uses the predicted pseudo label to select target-related features over all network to obtain a more confident and better result. Such selectivity is formulated as an optimization problem to maximize the category score of each node. In addition, the greedy optimization strategy is utilized to solve this problem and we have mathematically proven that the proposed mechanism can reach the local optimum by mathematical induction. Note that such mechanism is flexible and can be introduced to most GCN-based model. Extensive experimental results on benchmark datasets show that the proposed method can promote the confidence of the final target category and improve the performance of GCNs in the inference stage.

Graph convolutional network Node classification Under-confident Confidence correction mechanism

SCI ; EI

英语

通讯

Funda-mental Research Funds for the Provincial Universities of Zhejiang[GK229909299001-001] ; Zhejiang Provincial Natural Science Foundation of China["LR22F020001","LY22F020028","LDT23F02025F02"] ; Natural Science Foundation of China["62286082","62072147"]

Computer Science ; Engineering

Computer Science, Artificial Intelligence ; Engineering, Electrical & Electronic

WOS:001276406900001

ELSEVIER SCI LTD

ENGINEERING

Web of Science

1.Zhejiang Univ Technol, Coll Math Sci, Hangzhou 310023, Peoples R China
2.Southern Univ Sci & Technol, Dept Stat & Data Sci, Shenzhen 518055, Peoples R China
3.East China Normal Univ, Sch Data Sci & Engn, Shanghai 200333, Peoples R China
4.Hangzhou Dianzi Univ, Sch Comp Sci & Technol, Hangzhou 310018, Peoples R China
5.Zhejiang Zhongcai Pipes Sci & Technol Co Ltd, Xinchang 312500, Peoples R China
6.Hangzhou Dianzi Univ, Shangyu Inst Sci & Engn, Shangyu 312300, Peoples R China
7.China Acad Elect & Informat Technol CETC, Beijing 100041, Peoples R China

Yuan, Junqing,Guo, Huanlei,Zhou, Chenyi,et al. Confidence correction for trained graph convolutional networks[J]. PATTERN RECOGNITION,2024,156.

Yuan, Junqing.,Guo, Huanlei.,Zhou, Chenyi.,Ding, Jiajun.,Kuang, Zhenzhong.,...&Liu, Yuan.(2024).Confidence correction for trained graph convolutional networks.PATTERN RECOGNITION,156.

Yuan, Junqing,et al."Confidence correction for trained graph convolutional networks".PATTERN RECOGNITION 156(2024).