Exploring Applications of State Space Models and Advanced Training Techniques in Sequential Recommendations: A Comparative Study on Efficiency and Performance

TL;DR

This paper compares different approaches to sequential recommendations, emphasizing efficiency and performance, by exploring state space models, large language models, and adaptive training techniques.

Contribution

It evaluates the effectiveness of State Space Models and advanced training methods in improving speed and reducing costs in sequential recommendation systems.

Findings

State Space Models achieve SOTA results with lower latency and memory.

Large Language Models enhance recommendation quality via Monolithic Preference Optimization.

Adaptive algorithms significantly reduce training costs and time.

Abstract

Recommender systems aim to estimate the dynamically changing user preferences and sequential dependencies between historical user behaviour and metadata. Although transformer-based models have proven to be effective in sequential recommendations, their state growth is proportional to the length of the sequence that is being processed, which makes them expensive in terms of memory and inference costs. Our research focused on three promising directions in sequential recommendations: enhancing speed through the use of State Space Models (SSM), as they can achieve SOTA results in the sequential recommendations domain with lower latency, memory, and inference costs, as proposed by arXiv:2403.03900 improving the quality of recommendations with Large Language Models (LLMs) via Monolithic Preference Optimization without Reference Model (ORPO); and implementing adaptive batch- and step-size algorithms to reduce costs and accelerate training processes.