Hai Nguyen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Hai Nguyen has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Cong Tran
Posts and Telecommunications Institute of Technology
6 shared publications - Hieu Dao
Posts and Telecommunications Institute of Technology
3 shared publications - Tung Vu
Posts and Telecommunications Institute of Technology (PTIT), Hanoi, Vietnam
3 shared publications - Binh Duc Mai
Posts and Telecommunications Institute of Technology
1 shared publication - Cuong Pham
Posts and Telecommunications Institute of Technology, Hanoi, Vietnam
1 shared publication - Hanh Hoang
Vietnam Institute of Digital Transformation and Innovation
1 shared publication
Latest Publications
- 2 September 2026
- Published by Springer Nature
- p. 162-172
Synthesizing Cultural Heritage: An End-to-End System for Designing Jewelry with Vietnamese Hue Imperial Motifs
- 2 September 2026
- Published by Springer Nature
- p. 162-172
Synthesizing Cultural Heritage: An End-to-End System for Designing Jewelry with Vietnamese Hue Imperial Motifs
- Open Access
- 30 June 2026
- Published by Elsevier in Computers and Education: Artificial Intelligence
- Vol. 11
Emotion-aware classroom quality assessment leveraging IoT-based real-time student monitoring
- 30 June 2026
- Published by Elsevier in Computers and Education: Artificial Intelligence
- Vol. 11
Open AccessEmotion-aware classroom quality assessment leveraging IoT-based real-time student monitoring
- 23 June 2026
- Published by Elsevier in Computer Vision and Image Understanding
- Vol. 270
LiveNeRF: Efficient face replacement through Neural Radiance Fields integration
- 23 June 2026
- Published by Elsevier in Computer Vision and Image Understanding
- Vol. 270
LiveNeRF: Efficient face replacement through Neural Radiance Fields integration
- Open Access
- 18 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 76700-76721
MIST: A Multilingual Dataset and Benchmark for Fine-Grained Audio Inpainting Tampering Localization
Partial speech inpainting, replacing a few words within a genuine utterance via voice cloning to alter its meaning, is an emerging audio deepfake threat. Unlike fully synthesized speech, the manipulated content constitutes only 2–7% of the utterance, and existing benchmarks are limited to single-region tampering with utterance-level labels. To enable systematic study of this problem, we introduce MIST, a large-scale multilingual dataset spanning six languages with 1–3 independently inpainted word-level segments per utterance, generated through LLM-guided semantic replacement and neural voice cloning, totaling approximately 497k tampered utterances with precise temporal annotations. To establish a benchmark for this task, we further introduce ISA, a backbone-agnostic coarse-to-fine localization framework that recovers all tampered regions without prior knowledge of their count; and SF1@τ, a segment-level F1 metric based on temporal IoU matching that jointly evaluates region counting and boundary precision. Zero-shot experiments confirm that utterance-level classifiers trained on fully synthesized speech fail on MIST (SF1@0.5 ≤ 1.7% across all backbones), while fine-tuning on MIST annotations dramatically improves performance: the Wav2Vec2-AASIST backbone improves from 1.2% to 29.2%, and the strongest backbone (WavLM-AASIST) reaches SF1@0.5 = 33.5%, demonstrating that the primary bottleneck is training data rather than inference methodology. The dataset, code, and evaluation toolkit are publicly available.
- 18 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 76700-76721
Open AccessMIST: A Multilingual Dataset and Benchmark for Fine-Grained Audio Inpainting Tampering Localization
Partial speech inpainting, replacing a few words within a genuine utterance via voice cloning to alter its meaning, is an emerging audio deepfake threat. Unlike fully synthesized speech, the manipulated content constitutes only 2–7% of the utterance, and existing benchmarks are limited to single-region tampering with utterance-level labels. To enable systematic study of this problem, we introduce MIST, a large-scale multilingual dataset spanning six languages with 1–3 independently inpainted word-level segments per utterance, generated through LLM-guided semantic replacement and neural voice cloning, totaling approximately 497k tampered utterances with precise temporal annotations. To establish a benchmark for this task, we further introduce ISA, a backbone-agnostic coarse-to-fine localization framework that recovers all tampered regions without prior knowledge of their count; and SF1@τ, a segment-level F1 metric based on temporal IoU matching that jointly evaluates region counting and boundary precision. Zero-shot experiments confirm that utterance-level classifiers trained on fully synthesized speech fail on MIST (SF1@0.5 ≤ 1.7% across all backbones), while fine-tuning on MIST annotations dramatically improves performance: the Wav2Vec2-AASIST backbone improves from 1.2% to 29.2%, and the strongest backbone (WavLM-AASIST) reaches SF1@0.5 = 33.5%, demonstrating that the primary bottleneck is training data rather than inference methodology. The dataset, code, and evaluation toolkit are publicly available.
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 1
CounterFair: Mitigating Demographic Bias through StyleGAN3-Based Counterfactual Data Augmentation
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 1
Open AccessCounterFair: Mitigating Demographic Bias through StyleGAN3-Based Counterfactual Data Augmentation
- 26 April 2025
- Published by Springer Nature
- p. 246-256
Real-Time Multi-face Emotion Recognition in Classroom
- 26 April 2025
- Published by Springer Nature
- p. 246-256
Real-Time Multi-face Emotion Recognition in Classroom