Approximate Nearest Neighbor Negative Contrastive Learning for Dense Text Retrieval
This addresses the bottleneck in dense retrieval training for text retrieval applications, offering a significant performance boost and speed improvement, though it is incremental in refining existing methods.
The paper tackled the problem of dense text retrieval's reliance on sparse retrieval by addressing the discrepancy between training and testing data distributions, resulting in ANCE, which improved BERT-Siamese DR model to outperform competitive baselines and nearly match sparse-retrieval-and-BERT-reranking accuracy with almost 100x speed-up.
Conducting text retrieval in a dense learned representation space has many intriguing advantages over sparse retrieval. Yet the effectiveness of dense retrieval (DR) often requires combination with sparse retrieval. In this paper, we identify that the main bottleneck is in the training mechanisms, where the negative instances used in training are not representative of the irrelevant documents in testing. This paper presents Approximate nearest neighbor Negative Contrastive Estimation (ANCE), a training mechanism that constructs negatives from an Approximate Nearest Neighbor (ANN) index of the corpus, which is parallelly updated with the learning process to select more realistic negative training instances. This fundamentally resolves the discrepancy between the data distribution used in the training and testing of DR. In our experiments, ANCE boosts the BERT-Siamese DR model to outperform all competitive dense and sparse retrieval baselines. It nearly matches the accuracy of sparse-retrieval-and-BERT-reranking using dot-product in the ANCE-learned representation space and provides almost 100x speed-up.