RECE: Reduced Cross-Entropy Loss for Large-Catalogue Sequential Recommenders
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909286915899392 |
|---|---|
| author | Gusak, Danil Mezentsev, Gleb Oseledets, Ivan Frolov, Evgeny |
| author_facet | Gusak, Danil Mezentsev, Gleb Oseledets, Ivan Frolov, Evgeny |
| contents | Scalability is a major challenge in modern recommender systems. In sequential recommendations, full Cross-Entropy (CE) loss achieves state-of-the-art recommendation quality but consumes excessive GPU memory with large item catalogs, limiting its practicality. Using a GPU-efficient locality-sensitive hashing-like algorithm for approximating large tensor of logits, this paper introduces a novel RECE (REduced Cross-Entropy) loss. RECE significantly reduces memory consumption while allowing one to enjoy the state-of-the-art performance of full CE loss. Experimental results on various datasets show that RECE cuts training peak memory usage by up to 12 times compared to existing methods while retaining or exceeding performance metrics of CE loss. The approach also opens up new possibilities for large-scale applications in other domains. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_02354 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | RECE: Reduced Cross-Entropy Loss for Large-Catalogue Sequential Recommenders Gusak, Danil Mezentsev, Gleb Oseledets, Ivan Frolov, Evgeny Information Retrieval Machine Learning Scalability is a major challenge in modern recommender systems. In sequential recommendations, full Cross-Entropy (CE) loss achieves state-of-the-art recommendation quality but consumes excessive GPU memory with large item catalogs, limiting its practicality. Using a GPU-efficient locality-sensitive hashing-like algorithm for approximating large tensor of logits, this paper introduces a novel RECE (REduced Cross-Entropy) loss. RECE significantly reduces memory consumption while allowing one to enjoy the state-of-the-art performance of full CE loss. Experimental results on various datasets show that RECE cuts training peak memory usage by up to 12 times compared to existing methods while retaining or exceeding performance metrics of CE loss. The approach also opens up new possibilities for large-scale applications in other domains. |
| title | RECE: Reduced Cross-Entropy Loss for Large-Catalogue Sequential Recommenders |
| topic | Information Retrieval Machine Learning |
| url | https://arxiv.org/abs/2408.02354 |