Jointly Optimizing Query Encoder and Product Quantization to Improve Retrieval Performance

By Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Jiafeng Guo, M. Zhang, Shaoping Ma. 2021

97 citations recorded by the source corpus.

Recently, Information Retrieval community has witnessed fast-paced advances in Dense Retrieval (DR), which performs first-stage retrieval with embedding-based search. Despite the impressive ranking performance, previous studies usually adopt brute-force search to acquire candidates, which is prohibitive in practical Web search scenarios due to its tremendous memory usage and time cost. To overcome these problems, vector compression methods have been adopted in many practical embedding-based retrieval applications. One of the most popular methods is Product Quantization (PQ). However, although existing vector compression methods including PQ can help improve the efficiency of DR, they incur severely decayed retrieval performance due to the separation between encoding and compression. To tackle this problem, we present JPQ, which stands for Joint optimization of query encoding and Product Quantization. It trains the query encoder and PQ index jointly in an end-to-end manner based on three optimization strategies, namely ranking-oriented loss, PQ centroid optimization, and end-to-end negative sampling. We evaluate JPQ on two publicly available retrieval benchmarks. Experimental results show that JPQ significantly outperforms popular vector compression methods. Compared with previous DR models that use brute-force search, JPQ almost matches the best retrieval performance with 30x compression on index size. The compressed index further brings 10x speedup on CPU and 2x speedup on GPU in query latency.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Information Retrieval & Search

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Information Retrieval & Search

Citations in the corpus

The Web Is Your Oyster - Knowledge-Intensive NLP against a Very Large Web Corpus (2021, methodology): “phisticated bi-encoder training pipelines (Yamada et al., 2021; Zhan et al., 2021).”
Dense Text Retrieval Based on Pretrained Language Models: A Survey (2022, methodology): “[117] propose an end-to-end learning approach based on product quantization for dense retrieval, where three major optimization strategies are proposed, including rankingoriented loss, centroid optimization, and end-to-end negative sampling.”
Recommendation as Instruction Following: A Large Language Model Empowered Recommendation Approach (2023, unknown): “An example of instruction and its reversed version is shown below: [Normal Instruction]: “The user searches that <Query>, can you generate the item to respond to his query?””
TU Wien at TREC DL and Podcast 2021: Simple Compression for Dense Retrieval (2021, unknown): “…× , where 𝑏 is the output dimension of BERT and 𝑐 is our target compression dimension: Concurrent related works have also tackled the output compression of dense retrieval models: Zhan et al. [26] created a training procedure to optimize the product quantization of dense retrieval output vectors.”
Boosted Dense Retriever (2021, background): “The success of Dense Retrieval has led to many recent papers proposing schemes to improve dense retriever training by in-novating on how negatives are sampled (Xiong et al., 2020; Qu et al., 2021; Zhan et al., 2021c; Lin et al., 2021, inter alia.)”
Introducing Neural Bag of Whole-Words with ColBERTer: Contextualized Late Interactions using Enhanced Reduction (2022, methodology): “[58] optimize product quantization as part of the training.”
ColBERTv2: Effective and Efficient Retrieval via Lightweight Late Interaction (2021, methodology): “JPQ (Zhan et al., 2021a) and its extension, RepCONC (Zhan et al., 2021b), use product quantization (PQ) to compress embeddings, and jointly train the query encoder along with the centroids produced by PQ via a ranking-oriented loss function.”
When Is 0.1% Enough? Analyzing the Combined Effects of Dimensionality Reduction and Quantization on Text Embedding Compression (2026, unknown): “Isotropic Iterative Quantization compresses word embeddings into binary vectors (Liao et al., 2020), while product quantization has been used to reduce the memory and search cost of dense retrieval indexes (Zhan et al., 2021; Qiu et al., 2022).”