Ming Jiang
Ming Jiang has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
Huiteng Sun
2 shared publicationsMeiqu Lu
2 shared publications- Lei Tian
School of Artificial Intelligence, Guangxi Minzu University, Nanning, Guangxi, China
1 shared publication - Lina Ge
School of Artificial Intelligence, Guangxi Minzu University, Nanning, Guangxi, China and Guangxi Key Laboratory of Hybrid Computation and IC Design Analysis, Nanning, Guangxi, China
1 shared publication - Qinglin Bi
School of Artificial Intelligence, Guangxi Minzu University, Nanning, Guangxi, China
1 shared publication - Wenbo Lin
School of Artificial Intelligence, Guangxi University for Nationalities, No.188 Daxue East Road, Nanning, Guangxi 530006, China
1 shared publication
Latest Publications
- 25 September 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 352
ViSA: Subquadratic cross-modal attention with dual-path subtractive fusion
- 25 September 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 352
ViSA: Subquadratic cross-modal attention with dual-path subtractive fusion
- Open Access
- 8 May 2026
- Published by Elsevier in SSRN Electronic Journal
ViSA: Sub-Quadratic Cross-Modal Attention with Dual-Path Subtractive Fusion
Video-text retrieval becomes increasingly expensive when the number of videos grows, because standard cross-attention scales quadratically with sequence length. Existing methods usually improve either retrieval accuracy or computational efffciency, but rarely both at the same time. To address this issue, we present ViSA(Video-text interactive Synergistic Attention), a video-text interaction framework based on kernelized linear attention and subtraction-based fusion. The model reduces the cost of crossmodal interaction by pre-aggregating video features, so the overall computation grows linearly with the text and video lengths instead of forming a full attention matrix. At the same time, the differential design helps preserve useful discriminative information that is often weakened in linear attention. We evaluate ViSA on four benchmarks: MSR-VTT, LSMDC, DiDeMo, and VATEX. The results demonstrate that ViSA establishes a new state-of-the-art, outperforming strong CLIP-based baselines in accuracy while consistently using fewer GFLOPs. On MSR-VTT, it reaches 51.3% R@1 with 58 GFLOPs. Overall, ViSA provides an efffcient way to handle large-scale video-text retrieval without relying on quadratic softmax attention.
- 8 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessViSA: Sub-Quadratic Cross-Modal Attention with Dual-Path Subtractive Fusion
Video-text retrieval becomes increasingly expensive when the number of videos grows, because standard cross-attention scales quadratically with sequence length. Existing methods usually improve either retrieval accuracy or computational efffciency, but rarely both at the same time. To address this issue, we present ViSA(Video-text interactive Synergistic Attention), a video-text interaction framework based on kernelized linear attention and subtraction-based fusion. The model reduces the cost of crossmodal interaction by pre-aggregating video features, so the overall computation grows linearly with the text and video lengths instead of forming a full attention matrix. At the same time, the differential design helps preserve useful discriminative information that is often weakened in linear attention. We evaluate ViSA on four benchmarks: MSR-VTT, LSMDC, DiDeMo, and VATEX. The results demonstrate that ViSA establishes a new state-of-the-art, outperforming strong CLIP-based baselines in accuracy while consistently using fewer GFLOPs. On MSR-VTT, it reaches 51.3% R@1 with 58 GFLOPs. Overall, ViSA provides an efffcient way to handle large-scale video-text retrieval without relying on quadratic softmax attention.
- 10 January 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 336
HFLMND: Toward robust and efficient hierarchical federated learning via malicious node detection
- 10 January 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 336
HFLMND: Toward robust and efficient hierarchical federated learning via malicious node detection