Learnable Model Augmentation Contrastive Learning for Sequential Recommendation

• Published in: IEEE transactions on knowledge and data engineering Vol. 36; no. 8; pp. 3963 - 3976
• Main Authors: Hao, Yongjing, Zhao, Pengpeng, Xian, Xuefeng, Liu, Guanfeng, Zhao, Lei, Liu, Yanchi, Sheng, Victor S., Zhou, Xiaofang
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.08.2024; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Contrastive learning; Data augmentation; Data models; Electronic mail; learnable dropout; Learning; Markov processes; model augmentation; multi-positive pairs; Neurons; Recommender systems; Semantics; sequential recommendation; Sparsity; Task analysis
• ISSN: 1041-4347; 1558-2191
• DOI: 10.1109/TKDE.2023.3330426

Sequential Recommendation (SR) methods play a crucial role in recommender systems, which aims to capture users' dynamic interest from their historical interactions. Recently, Contrastive Learning (CL), which has emerged as a successful method for sequential recommendation, utilizes various data augmentations to generate contrastive views to mine supervised signals from data to alleviate data sparsity issues. However, most existing sequential data augmentation methods may destroy semantic sequential interaction characteristics. Meanwhile, they often adopt random operations when generating contrastive views leading to suboptimal performance. To this end, in this paper, we propose a L earnable M odel A ugmentation Contrastive learning for sequential Rec ommendation (LMA4Rec) . Specifically, LMA4Rec first takes the model-based augmentation method to generate constructive views. Then, LMA4Rec uses Learnable Bernoulli Dropout (LBD) to implement learnable model augmentation operations. Next, contrastive learning is used between the contrastive views to extract supervised signals. Furthermore, a novel multi-positive contrastive learning loss alleviates the supervised sparsity issue. Finally, experiments on public datasets show that our LMA4Rec method effectively improved sequential recommendation performance compared with the state-of-the-art baseline methods.