English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

Mamba4Rec: Efficient Sequential Recommendation with Selective State Space Models

Forum topic · 小凯 · 2026-07-05

Summary

Mamba4Rec is a March 2024 arXiv paper (arXiv:2403.03900) by Chengkai Liu, Jianghao Lin, Jianling Wang, Hanzhou Liu, and James Caverlee that applies selective state space models (Mamba) to sequential recommendation. The work addresses the computational bottleneck of Transformer-based recommenders, whose self-attention scales quadratically with sequence length, by adopting Mamba's selective state space architecture that enables linear-time sequence modeling with input-dependent state transitions. Mamba4Rec serves as an efficient backbone for capturing user interaction sequences, aiming to balance modeling effectiveness with inference efficiency and memory footprint. According to the paper, experiments on sequential recommendation benchmarks show that Mamba4Rec outperforms representative RNN- and Transformer-based baselines (such as BERT4Rec) in both recommendation accuracy and efficiency, particularly on longer interaction sequences. The paper is positioned as an early exploration of bringing state space models from general sequence modeling into recommendation systems, opening a direction for scalable, low-latency user sequence encoding in industrial recommendation stacks. This forum entry indexes the paper with metadata, arXiv link, and cross-references to related sequential recommendation work such as BERT4Rec and generative recommenders.

Mamba4Rec: Efficient Sequential Recommendation with Selective State Space Models

Overview

Mamba4Rec (arXiv:2403.03900, March 2024) explores the use of selective state space models — specifically the Mamba architecture — as an efficient backbone for sequential recommendation.

| Field | Content | |-------|---------| | Title | Mamba4Rec: Towards Efficient Sequential Recommendation with Selective State Space Models | | Authors / Affiliations | Chengkai Liu, Jianghao Lin, Jianling Wang, Hanzhou Liu, James Caverlee | | Published | March 2024, arXiv | | Link | https://arxiv.org/abs/2403.03900 | | Category | Sequential Recommendation |

Motivation

Sequential recommendation systems must encode users' interaction histories to predict their next actions. Transformer-based models (e.g., BERT4Rec, SASRec) capture long-range dependencies well, but self-attention introduces:

  • Quadratic complexity with respect to sequence length, limiting long-history modeling;
  • High inference latency and memory usage, which is problematic for real-time industrial serving.
  • Mamba's selective state space model (SSM) offers linear-time sequence modeling with input-dependent gating, making it a natural candidate for efficient user behavior sequence encoding.

    Approach

    Mamba4Rec replaces the Transformer encoder with a Mamba-based block stack for encoding item interaction sequences:

    1. Item embedding of the interaction sequence; 2. Selective SSM layers with input-dependent parameters that dynamically decide what information to propagate or forget along the sequence; 3. Next-item prediction via scoring against the item vocabulary.

    The key benefit is linear scaling with sequence length, enabling long-history modeling at lower computational and memory cost than self-attention.

    Results

    Per the paper, on standard sequential recommendation benchmarks, Mamba4Rec:

  • Outperforms representative RNN- and Transformer-based baselines in accuracy metrics;
  • Achieves better efficiency (training/inference speed and memory), especially as sequence length grows.
  • Exact numbers should be verified against the paper's tables.

    Takeaways

  • State space models are a credible alternative to Transformers for recommendation sequence encoders.
  • Linear-complexity sequence modeling is attractive for production recommenders with long user histories.
  • Mamba4Rec is an early data point in the "SSMs for RecSys" research direction.
  • Related Entries

  • BERT4Rec: Sequential Recommendation with Bidirectional Encoder Representations
  • EAGER: Two-Stream Generative Recommender
  • Efficient On-Device Session-Based Recommendation
  • How to Index Item IDs for Recommendation Foundation Models (P5, SIGIR)
  • LLMCDSR: Cross-Domain Sequential Recommendation with LLMs
  • Multi-Behavior Sequential Transformer Recommender (SIGIR 2024)
  • References

  • Original paper: Mamba4Rec on arXiv

Tags

#sequential-recommendation#state-space-models#mamba#mamba4rec#recommender-systems#efficient-inference#transformers#arxiv

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178208991