More Than Efficiency: Embedding Compression Improves Domain Adaptation in Dense Retrieval
For practitioners of dense retrieval, this provides a simple, annotation-free domain adaptation method that outperforms no adaptation in most cases.
The paper shows that applying PCA to compress query embeddings improves dense retrieval performance in specialized domains, achieving NDCG@10 gains in 75.4% of 126 model-dataset pairs across 9 retrievers and 14 datasets.
Dense retrievers powered by pretrained embeddings are widely used for document retrieval but struggle in specialized domains due to the mismatches between the training and target domain distributions. Domain adaptation typically requires costly annotation and retraining of query-document pairs. In this work, we revisit an overlooked alternative: applying PCA to domain embeddings to derive lower-dimensional representations that preserve domain-relevant features while discarding non-discriminative components. Though traditionally used for efficiency, we demonstrate that this simple embedding compression can effectively improve retrieval performance. Evaluated across 9 retrievers and 14 MTEB datasets, PCA applied solely to query embeddings improves NDCG@10 in 75.4% of model-dataset pairs, offering a simple and lightweight method for domain adaptation.