Sequential Recommendation via Adaptive Robust Attention with Multi-dimensional Embeddings

TL;DR

This paper introduces ADRRec, an advanced sequential recommendation model that enhances robustness and accuracy by integrating adaptive robust attention, multi-dimensional embeddings, and layer-wise noise injection, outperforming existing architectures.

Contribution

It proposes a novel adaptive robust attention mechanism with multi-dimensional embeddings and layer-wise noise injection for improved sequential recommendation.

Findings

Outperforms existing self-attention models in accuracy.

Demonstrates robustness and generalization improvements.

Achieves superior results across extensive experiments.

Abstract

Sequential recommendation models have achieved state-of-the-art performance using self-attention mechanism. It has since been found that moving beyond only using item ID and positional embeddings leads to a significant accuracy boost when predicting the next item. In recent literature, it was reported that a multi-dimensional kernel embedding with temporal contextual kernels to capture users' diverse behavioral patterns results in a substantial performance improvement. In this study, we further improve the sequential recommender model's robustness and generalization by introducing a mix-attention mechanism with a layer-wise noise injection (LNI) regularization. We refer to our proposed model as adaptive robust sequential recommendation framework (ADRRec), and demonstrate through extensive experiments that our model outperforms existing self-attention architectures.