Mingyan Wu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Mingyan Wu has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
- Boer Zhang
School of Engineering and Applied Sciences, Harvard University, Cambridge, Massachusetts
1 shared publication - Dongzhuoran Zhou
University of Oslo, Oslo, Norway and Bosch Center for AI, Baden-Württemberg, Germany
1 shared publication - Fu
School of Mechanical Engineering, Yangtze University, Jinzhou, China
1 shared publication - Guohao Li
Oxford University, Oxford, United Kingdom
1 shared publication - Han
School of Electronic Information, Yangtze University, Jinzhou, China
1 shared publication - Mu
School of Mechanical Engineering, Yangtze University, Jinzhou, China
1 shared publication
Latest Publications
- Open Access
- 26 May 2026
- Published by Association for Computing Machinery (ACM)
- p. 1114-1118
EigentSearch-Q+: Enhancing Deep Research Agents with Structured Reasoning Tools
Deep research requires reasoning over web evidence to answer open-ended questions, and it is a core capability for AI agents. Yet many deep research agents still rely on implicit, unstructured search behavior that causes redundant exploration and brittle evidence aggregation. Motivated by Anthropic’s “think” tool paradigm and insights from the information-retrieval literature, we introduce Q+, a set of query and evidence processing tools that make web search more deliberate by guiding query planning, monitoring search progress, and extracting evidence from long web snapshots. We integrate Q+ into the browser sub-agent of Eigent, an open-source, production-ready multi-agent workforce for computer use, yielding EigentSearch-Q+. Across four benchmarks (SimpleQA-Verified, FRAMES, WebWalkerQA, and XBench DeepSearch), Q+ improves Eigent’s browser agent benchmark-size-weighted average accuracy by 3.0, 3.8, and 0.6 percentage points (pp) for GPT-4.1, GPT-5.1, and Minimax M2.5 model backends, respectively. Case studies further suggest that EigentSearch-Q+ produces more coherent tool-calling trajectories by making search progress and evidence handling explicit.1
- 26 May 2026
- Published by Association for Computing Machinery (ACM)
- p. 1114-1118
Open AccessEigentSearch-Q+: Enhancing Deep Research Agents with Structured Reasoning Tools
Deep research requires reasoning over web evidence to answer open-ended questions, and it is a core capability for AI agents. Yet many deep research agents still rely on implicit, unstructured search behavior that causes redundant exploration and brittle evidence aggregation. Motivated by Anthropic’s “think” tool paradigm and insights from the information-retrieval literature, we introduce Q+, a set of query and evidence processing tools that make web search more deliberate by guiding query planning, monitoring search progress, and extracting evidence from long web snapshots. We integrate Q+ into the browser sub-agent of Eigent, an open-source, production-ready multi-agent workforce for computer use, yielding EigentSearch-Q+. Across four benchmarks (SimpleQA-Verified, FRAMES, WebWalkerQA, and XBench DeepSearch), Q+ improves Eigent’s browser agent benchmark-size-weighted average accuracy by 3.0, 3.8, and 0.6 percentage points (pp) for GPT-4.1, GPT-5.1, and Minimax M2.5 model backends, respectively. Case studies further suggest that EigentSearch-Q+ produces more coherent tool-calling trajectories by making search progress and evidence handling explicit.1
- Open Access
- 9 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 112085-112096
Remaining Useful Life Prediction via a Data-Driven Deep Learning Fusion Model-CALAP
As one of the key technologies in the field of Prognostic and Health Management (PHM), Remaining Useful Life (RUL) prediction technology plays an important role in equipment health maintenance and fault detection. For complex devices, the degradation process of the remaining useful life of the device is often difficult to be described with mathematical or physical models, so data-driven methods has become an important and feasible method in the field of RUL prediction. This paper proposes a data-driven deep learning fusion model based on CNN-ATTENTION-LSTM-ATTENTION-PARALLEL (CALAP) model, in which Convolutional Neural Networks (CNN) and Long Short-Term Memory Networks (LSTM) are adopted to extract the spatial features and temporal features of the data in parallel, and both of them combine the corresponding attention mechanism allowing the network to focus on important factors. The CNN path fuses CBAM and the LSTM path fuses attention mechanism. We evaluate the proposed model on the C-MAPSS dataset released by NASA and compare it to the state-of-the-art. The RMSE of the proposed model is reduced by nearly 2-5%, and the score is reduced by nearly 8-10% under simple conditions. Experimental results prove that the model has relatively high prediction accuracy and good robustness.
- 9 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 112085-112096
Open AccessRemaining Useful Life Prediction via a Data-Driven Deep Learning Fusion Model-CALAP
As one of the key technologies in the field of Prognostic and Health Management (PHM), Remaining Useful Life (RUL) prediction technology plays an important role in equipment health maintenance and fault detection. For complex devices, the degradation process of the remaining useful life of the device is often difficult to be described with mathematical or physical models, so data-driven methods has become an important and feasible method in the field of RUL prediction. This paper proposes a data-driven deep learning fusion model based on CNN-ATTENTION-LSTM-ATTENTION-PARALLEL (CALAP) model, in which Convolutional Neural Networks (CNN) and Long Short-Term Memory Networks (LSTM) are adopted to extract the spatial features and temporal features of the data in parallel, and both of them combine the corresponding attention mechanism allowing the network to focus on important factors. The CNN path fuses CBAM and the LSTM path fuses attention mechanism. We evaluate the proposed model on the C-MAPSS dataset released by NASA and compare it to the state-of-the-art. The RMSE of the proposed model is reduced by nearly 2-5%, and the score is reduced by nearly 8-10% under simple conditions. Experimental results prove that the model has relatively high prediction accuracy and good robustness.
- Open Access
- 15 March 1998
- Published by EDP Sciences in Astronomy and Astrophysics Supplement Series
- Vol. 128 (2) , 363-368
Direct demodulation technique for rotating modulation collimator imaging
Astronomy and Astrophysics, Supplement Series (A&AS) published data papers, either observational or theoretical, as well as extensive data material forming the basis of papers with astrophysical results
- 15 March 1998
- Published by EDP Sciences in Astronomy and Astrophysics Supplement Series
- Vol. 128 (2) , 363-368
Open AccessDirect demodulation technique for rotating modulation collimator imaging
Astronomy and Astrophysics, Supplement Series (A&AS) published data papers, either observational or theoretical, as well as extensive data material forming the basis of papers with astrophysical results