Tuan Dang-Minh

Tuan Dang-Minh

Auto-generated profile, based on publicly available information

Tuan Dang-Minh

CMC Applied Technology Institute, CMC Corporation

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
20 August 2023
Verified
Book Chapter
Closed access
PDF file
Xuan-Ha NguyenXuan-Ha Nguyen
Xuan-Ha Nguyen
DHDong Nhu Hoang
Dong Nhu Hoang
TDTuan Minh Dang
Tuan Minh Dang


5 July 2022
Verified
Book Chapter
Closed access
PDF file
Xuan-Ha NguyenXuan-Ha Nguyen
Xuan-Ha Nguyen
DHDong Hoang-Nhu
Dong Hoang-Nhu
ANAnh Nguyen-Duy
Anh Nguyen-Duy
TDTuan Dang-Minh
Tuan Dang-Minh

20 August 2023
Verified
Book Chapter
Closed access
PDF file
Xuan-Ha NguyenXuan-Ha Nguyen
Xuan-Ha Nguyen
DHDong Nhu Hoang
Dong Nhu Hoang
HBHoang Viet Bui
Hoang Viet Bui
TDTuan Minh Dang
Tuan Minh Dang

5 July 2022
Verified
Book Chapter
Closed access
PDF file
Xuan-Ha NguyenXuan-Ha Nguyen
Xuan-Ha Nguyen
DHDong Hoang-Nhu
Dong Hoang-Nhu
KKKien Kim-Van
Kien Kim-Van
TDTuan Dang-Minh
Tuan Dang-Minh

7 December 2023
Verified
Conference Paper
Open access
LVLoi Le Vu
Loi Le Vu
DTDong Nguyen Tien
Dong Nguyen Tien
TMTuan Dang Minh
Tuan Dang Minh
Dense passage retrieval has recently boosted the performance of involved systems such as question answering or search engine. On this problem, prior works trained a dense retriever by learning to rank, i.e. ranking relevant/positive passages higher than irrelevant/negative ones. In this paper, we propose a Stratified Ranking approach for Dense Passage Retrieval (SR-DPR), which performs three-way ranking instead of the typical two-way positive-negative ranking. SR-DPR is concerned with three relevance levels, which are positive, hard negative and random/easy negative. We train SR-DPR model by minimizing a contrastive negative log-likelihood (NLL) loss involved with these three relevance levels, which is a finer-grained version of the N-pair loss [16]. To efficiently implement SR-DPR, we designed three data pipelines, each pipeline is used to learn the contrast between two out of three relevance levels. SR-DPR outperforms the strong baseline DPR [5] by 0.6-1.5% retrieval accuracy on Natural questions [6] dataset and 3-6% of that on Zalo Legal Text Retrieval dataset. SR-DPR also gives competitive results compared with current state-of-the-art methods without requiring complicated training regime or intensive hardware resources. The idea of stratified ranking of SR-DPR is not restricted to the scope of dense passage retrieval but can be applied in any contrastive learning problem. We conducted detailed ablation studies to give insights into SR-DPR’s behavior.