The Future is Sparse: Embedding Compression for Scalable Retrieval in Recommender Systems
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913842412388352 |
|---|---|
| author | Kasalický, Petr Spišák, Martin Vančura, Vojtěch Bohuněk, Daniel Alves, Rodrigo Kordík, Pavel |
| author_facet | Kasalický, Petr Spišák, Martin Vančura, Vojtěch Bohuněk, Daniel Alves, Rodrigo Kordík, Pavel |
| contents | Industry-scale recommender systems face a core challenge: representing entities with high cardinality, such as users or items, using dense embeddings that must be accessible during both training and inference. However, as embedding sizes grow, memory constraints make storage and access increasingly difficult. We describe a lightweight, learnable embedding compression technique that projects dense embeddings into a high-dimensional, sparsely activated space. Designed for retrieval tasks, our method reduces memory requirements while preserving retrieval performance, enabling scalable deployment under strict resource constraints. Our results demonstrate that leveraging sparsity is a promising approach for improving the efficiency of large-scale recommenders. We release our code at https://github.com/recombee/CompresSAE. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_11388 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | The Future is Sparse: Embedding Compression for Scalable Retrieval in Recommender Systems Kasalický, Petr Spišák, Martin Vančura, Vojtěch Bohuněk, Daniel Alves, Rodrigo Kordík, Pavel Information Retrieval Machine Learning Industry-scale recommender systems face a core challenge: representing entities with high cardinality, such as users or items, using dense embeddings that must be accessible during both training and inference. However, as embedding sizes grow, memory constraints make storage and access increasingly difficult. We describe a lightweight, learnable embedding compression technique that projects dense embeddings into a high-dimensional, sparsely activated space. Designed for retrieval tasks, our method reduces memory requirements while preserving retrieval performance, enabling scalable deployment under strict resource constraints. Our results demonstrate that leveraging sparsity is a promising approach for improving the efficiency of large-scale recommenders. We release our code at https://github.com/recombee/CompresSAE. |
| title | The Future is Sparse: Embedding Compression for Scalable Retrieval in Recommender Systems |
| topic | Information Retrieval Machine Learning |
| url | https://arxiv.org/abs/2505.11388 |