TV-Rec: Time-Variant Convolutional Filter for Sequential Recommendation

2025-10-29

AI Summary

Existing sequential recommendation methods face two key limitations: fixed convolutional kernels struggle to capture global temporal interactions, while self-attention mechanisms incur high computational overhead. To address these issues, this paper proposes Time-Varying Component Convolution (TVC), the first approach to incorporate graph signal processing principles into convolutional filter design. TVC employs a dynamic kernel generation mechanism to adaptively model temporal dependencies and positional dynamics within user behavior sequences. By replacing both conventional fixed convolutions and self-attention modules, TVC enhances representational capacity while significantly reducing computational cost. Extensive experiments on six public benchmark datasets demonstrate that the proposed method achieves an average 7.49% improvement in recommendation accuracy over state-of-the-art approaches, striking an effective balance between predictive performance and inference efficiency.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Search and Optimization: Non-convex Optimization

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Recently, convolutional filters have been increasingly adopted in sequential recommendation for their ability to capture local sequential patterns. However, most of these models complement convolutional filters with self-attention. This is because convolutional filters alone, generally fixed filters, struggle to capture global interactions necessary for accurate recommendation. We propose Time-Variant Convolutional Filters for Sequential Recommendation (TV-Rec), a model inspired by graph signal processing, where time-variant graph filters capture position-dependent temporal variations in user sequences. By replacing both fixed kernels and self-attention with time-variant filters, TV-Rec achieves higher expressive power and better captures complex interaction patterns in user behavior. This design not only eliminates the need for self-attention but also reduces computation while accelerating inference. Extensive experiments on six public benchmarks show that TV-Rec outperforms state-of-the-art baselines by an average of 7.49%.

Problem

• Replacing fixed filters and self-attention with time-variant convolutional filters
• Capturing position-dependent temporal variations in user sequences
• Improving sequential recommendation accuracy while reducing computation

Innovation

• Time-variant filters replace fixed convolutional kernels
• Time-variant filters eliminate self-attention mechanism
• Graph signal processing captures position-dependent temporal variations