Learning to Embed Categorical Features without Embedding Tables for Recommendation
Abstract
Embedding learning of categorical features (e.g. user/item IDs) is at the core of various recommendation models including matrix factorization and neural collaborative filtering. The standard approach creates an embedding table where each row represents a dedicated embedding vector for every unique feature value. However, this method fails to efficiently handle high-cardinality features and unseen feature values (e.g. new video ID) that are prevalent in real-world recommendation systems. In this paper, we propose an alternative embedding framework Deep Hash Embedding (DHE), replacing embedding tables by a deep embedding network to compute embeddings on the fly. DHE first encodes the feature value to a unique identifier vector with multiple hashing functions and transformations, and then applies a DNN to convert the identifier vector to an embedding. The encoding module is deterministic, non-learnable, and free of storage, while the embedding network is updated during the training time to learn embedding generation. Empirical results show that DHE achieves comparable AUC against the standard one-hot full embedding, with smaller model sizes. Our work sheds light on the design of DNN-based alternative embedding schemes for categorical features without using embedding table lookup.
Cite
@article{arxiv.2010.10784,
title = {Learning to Embed Categorical Features without Embedding Tables for Recommendation},
author = {Wang-Cheng Kang and Derek Zhiyuan Cheng and Tiansheng Yao and Xinyang Yi and Ting Chen and Lichan Hong and Ed H. Chi},
journal= {arXiv preprint arXiv:2010.10784},
year = {2021}
}
Comments
Accepted to KDD'21, Research Track