A5082461400- Topic Modeling
- Advanced Graph Neural Networks
- Multimodal Machine Learning Applications
- Natural Language Processing Techniques
- Data Quality and Management
- Video Analysis and Summarization
- Advanced Text Analysis Techniques
- Cancer-related molecular mechanisms research
- Advanced Image and Video Retrieval Techniques
- Human Pose and Action Recognition
- Cognitive Computing and Networks
- Medical Research and Treatments
- Advanced Decision-Making Techniques
- Software System Performance and Reliability
- bioluminescence and chemiluminescence research
- Luminescence Properties of Advanced Materials
- Video Surveillance and Tracking Methods
- Luminescence and Fluorescent Materials
- Bayesian Modeling and Causal Inference
- Educational Reforms and Innovations
- Image Retrieval and Classification Techniques
- Systems Engineering Methodologies and Applications
- Software Reliability and Analysis Research
- Semantic Web and Ontologies
- Military Strategy and Technology
Zhejiang Sci-Tech University
2025
Ludwig-Maximilians-Universität München
2023-2024
LMU Klinikum
2023
Munich Center for Machine Learning
2023
National University of Defense Technology
2012-2018
Prompt engineering is a technique that involves augmenting large pre-trained model with task-specific hints, known as prompts, to adapt the new tasks. Prompts can be created manually natural language instructions or generated automatically either vector representations. enables ability perform predictions based solely on prompts without updating parameters, and easier application of models in real-world In past years, has been well-studied processing. Recently, it also intensively studied...
Lanthanide-doped upconversion luminescent nanoparticles (UCNPs) have garnered extensive attention due to their notable anti-Stokes shifts and superior photostability. Notably, Ho3+-based UCNPs present a complex energy level configuration, which poses challenges in augmenting luminescence efficiency. Herein, rational design strategy was used enhance the intensity of Ho3+ ions by improving photon absorption ability utilization Efficient transfer excitation light were achieved through carefully...
Since conventional knowledge embedding models cannot take full advantage of the abundant textual information, there have been extensive research efforts in enhancing using texts. However, existing enhancement approaches apply to temporal graphs (tKGs), which contain time-dependent event with complex dynamics. Specifically, often assume is time-independent. In contrast, entity tKG usually evolves, poses challenge aligning temporally relevant texts entities. To this end, we propose study data...
Since conventional knowledge embedding models cannot take full advantage of the abundant textual information, there have been extensive research efforts in enhancing using texts. However, existing enhancement approaches apply to temporal graphs (tKGs), which contain time-dependent event with complex dynamics. Specifically, often assume is time-independent. In contrast, entity tKG usually evolves, poses challenge aligning temporally relevant texts entities. To this end, we propose study data...
We present EchoScene, an interactive and controllable generative model that generates 3D indoor scenes on scene graphs. EchoScene leverages a dual-branch diffusion dynamically adapts to Existing methods struggle handle graphs due varying numbers of nodes, multiple edge combinations, manipulator-induced node-edge operations. overcomes this by associating each node with denoising process enables collaborative information exchange, enhancing consistent generation aware global constraints. This...
In the video-language domain, recent works in leveraging zero-shot Large Language Model-based reasoning for video understanding have become competitive challengers to previous end-to-end models. However, long presents unique challenges due complexity of over extended timespans, even LLM-based approaches. The challenge information redundancy videos prompts question what specific is essential large language models (LLMs) and how leverage them complex spatial-temporal long-form analysis. We...
The rapid advancements in large language models (LLMs) have ignited interest the temporal knowledge graph (tKG) domain, where conventional embedding-based and rule-based methods dominate. question remains open of whether pre-trained LLMs can understand structured relational data replace them as foundation model for forecasting. Therefore, we bring forecasting into generative setting. However, challenges occur huge chasms between complex structure sequential natural expressions handle,...
While multi-modal models have successfully integrated information from image, video, and audio modalities, integrating graph modality into large language (LLMs) remains unexplored. This discrepancy largely stems the inherent divergence between structured data unstructured text data. Incorporating knowledge provides a reliable source of information, enabling potential solutions to address issues in generation, e.g., hallucination, lack domain knowledge. To evaluate integration models,...
Modeling evolving knowledge over temporal graphs (TKGs) has become a heated topic. Various methods have been proposed to forecast links on TKGs. Most of them are embedding-based, where hidden representations learned represent graph (KG) entities and relations based the observed contexts. Although these show strong performance traditional TKG forecasting (TKGF) benchmarks, they face challenge in modeling unseen zero-shot that no prior context. In this paper, we try mitigate problem as...
In this paper, the ability predicates are set up for system's functional unit, and influence rules built dependent relationship in system. With mathematical between indexes weights of predicates, factors with certain capability, uncertain performance, relationships, complex system integrated together. Based on Markov Logic Networks (MLNs) framework, effectiveness units, subsystems evaluated from units' layer, subsystems' layer to top systems' layer. The evaluation combined static entity's...
The complexity and the uncertainty of mission environment require accurate analysis evaluation for Joint course actions (COA). This paper proposed an approach complex COA, using influence analytical tools uncertainty, based on exploratory (EA) COA architecture. A multi-solution effectiveness model framework its were proposed, as well procedure it. case study a well-known landing operation was also carried out.
services and responsibilities on the library.Meanwhile, this thesis also provides some suggestions construction perfection of MOOC copyright.