A5101638182- Service-Oriented Architecture and Web Services
- Topic Modeling
- Advanced Computational Techniques and Applications
- Text and Document Classification Technologies
- Adaptive Control of Nonlinear Systems
- Natural Language Processing Techniques
- Remote Sensing and Land Use
- Remote-Sensing Image Classification
- Guidance and Control Systems
- Collaboration in agile enterprises
- Conducting polymers and applications
- Advanced Sensor and Energy Harvesting Materials
- Quantum and electron transport phenomena
- Traffic Prediction and Management Techniques
- Financial Markets and Investment Strategies
- Advanced Photocatalysis Techniques
- Advanced oxidation water treatment
- Stock Market Forecasting Methods
- Physics of Superconductivity and Magnetism
- Anomaly Detection Techniques and Applications
- Advanced Graph Neural Networks
- Advanced Image Fusion Techniques
- Advanced Sensor and Control Systems
- Sentiment Analysis and Opinion Mining
- Time Series Analysis and Forecasting
Zhejiang Shuren University
2010-2025
Stevens Institute of Technology
2024
Jiangsu Normal University
2024
Shanghai Institute of Microsystem and Information Technology
2020-2023
Chinese Academy of Sciences
2018-2023
Rutgers Sexual and Reproductive Health and Rights
2021-2023
University of Chinese Academy of Sciences
2020-2023
Princeton University
2022
Harbin Institute of Technology
2016-2021
Shenyang Aerospace University
2021
Recent advancements in Large Language Models (LLMs) have exhibited notable efficacy question-answering (QA) tasks across diverse domains. Their prowess integrating extensive web knowledge has fueled interest developing LLM-based autonomous agents. While LLMs are efficient decoding human instructions and deriving solutions by holistically processing historical inputs, transitioning to purpose-driven agents requires a supplementary rational architecture process multi-source information,...
Graph walking based on reinforcement learning (RL) has shown great success in navigating an agent to automatically complete various reasoning tasks over incomplete knowledge graph (KG) by exploring multi-hop relational paths. However, existing approaches only work well short paths and tend miss the target entity with increasing path length. This is undesirable for many real-world scenarios, where connecting source entities are not available KGs, thus performances drop drastically unless able...
Facial expression recognition plays a key role in human-computer emotional interaction. However, human faces real environments are affected by various unfavorable factors, which will result the reduction of accuracy. In this paper, we proposed novel method combines Fine-tuning Swin Transformer and Multiple Weights Optimality-seeking (FST-MWOS) to enhanced performance. FST-MWOS mainly consists two crucial components: (FST) (MWOS). FST takes Large as backbone network obtain multiple groups...
Large language models (LLMs) have demonstrated exceptional capabilities across a wide range of tasks but also pose significant risks due to their potential generate harmful content. Although existing safety mechanisms can improve model safety, they often lead overly cautious behavior and fail fully utilize LLMs' internal cognitive processes. Drawing inspiration from science, where humans rely on reflective reasoning (System 2 thinking) regulate behavior, we empirically demonstrate that LLMs...
Supercapacitors that can function when in direct contact with human tissue are of paramount importance for wearable bioelectronics but face mismatching biological and its movement. Herein, we developed a zwitterion hydrogel elastomer electrode-based all-hydrogel supercapacitor (AHSC) characterized by good energy storage properties, bioadhesion, body movement-matching mechanical biocompatibility. These functions were realized integrating...
Job Title Benchmarking (JTB) aims at matching job titles with similar expertise levels across various companies. JTB could provide precise guidance and considerable convenience for both talent recruitment seekers position salary calibration/prediction. Traditional approaches mainly rely on manual market surveys, which is expensive labor intensive. Recently, the rapid development of Online Professional graph has accumulated a large number career records, provides promising trend data-driven...
Promising progress has been made toward learning efficient time series representations in recent years, but the learned often lack interpretability and do not encode semantic meanings by complex interactions of many latent factors. Learning that disentangle these factors can bring semantic-rich further enhance interpretability. However, directly adopting sequential models, such as Long Short-Term Memory Variational AutoEncoder (LSTM-VAE), would encounter a Kullback?Leibler (KL) vanishing...
The essential task of urban planning is to generate the optimal land-use configuration a target area. However, traditional time-consuming and labor-intensive. Deep generative learning gives us hope that we can automate this process come up with ideal plans. While remarkable achievements have been obtained, they exhibited limitations in lacking awareness of: 1) hierarchical dependencies between functional zones spatial grids; 2) peer among zones; 3) human regulations ensure usability...
Multi-node wind speed forecasting is greatly important for offshore power. It a challenging task due to unknown complex spatial dependencies. Recently, graph neural networks (GNN) have been applied because of their capability in modeling However, existing methods usually require pre-defined structure, which not optimal the downstream and limits application scope GNN. In this paper, we propose adaptive graph-learning convolutional (AGLCN) that can automatically infer hidden associations among...
Adopting a capability-based view of digital transformation as 2nd-order ‘dynamic’ capability, this paper investigates how 1st-order dynamic and operational IT capabilities are strategically configured aligned by manufacturing SMEs in order to gain organizational agility. Resulting from fuzzy-set qualitative comparative analysis (fsQCA) 67 Canadian SMEs, our results show that high level agility is concretized when these firms align at least three one capability. Through high-performing...
Large language models (LLMs) have demonstrated notable potential in conducting complex tasks and are increasingly utilized various financial applications. However, high-quality sequential investment decision-making remains challenging. These require multiple interactions with a volatile environment for every decision, demanding sufficient intelligence to maximize returns manage risks. Although LLMs been used develop agent systems that surpass human teams yield impressive returns,...
Alzheimer’s Disease is a neurodegenerative disorder, and one of its common prominent early symptoms language impairment. Therefore, diagnosis through speech text information significant importance. However, the multimodal data often complex inconsistent, which leads to inadequate feature extraction. To address problem, We propose model for based on attention(EDAMM). Specifically, we first evaluate select three optimal extraction methods, Wav2Vec2.0, TF-IDF Word2Vec, extract acoustic...
Pre-trained language models such as BERT have achieved great success in a broad range of natural processing tasks. However, cannot well support E-commerce related tasks due to the lack two levels domain knowledge, i.e., phrase-level and product-level. On one hand, many require an accurate understanding phrases, whereas fine-grained knowledge is not explicitly modeled by BERT's training objective. other product-level like product associations can enhance modeling E-commerce, but they are...
Scalable memories that can match the speeds of superconducting logic circuits have long been desired to enable a computer. A loop includes Josephson junction store flux quantum state in picoseconds. However, requirement for inductance create bi-state hysteresis sets limit on minimal area occupied by single memory cell. Here, we present miniaturized cell based Three-Dimensional (3D) Nb nano-Superconducting QUantum Interference Device (nano-SQUID). The major here fits within an 8*9 {\mu}m^2...
Hierarchical Multi-Label Classification (HMLC) is a well-established problem that aims at assigning data instances to multiple classes stored in hierarchical structure. Despite its importance, existing approaches often face two key limitations: (i) They employ dense networks solely explore the class hierarchy as hard criterion for maintaining taxonomic consistency among predicted classes, yet without leveraging rich semantic relationships between and classes; (ii) struggle generalize...
The yellow mustard plant in Northern Shaanxi is a precious germplasm, and the seed trait controlled by single recessive gene. In this report, amplified fragment length polymorphism (AFLP) simple sequence repeat (SSR) techniques were used to identify markers linked brown locus an F(2) population consisting of 1258 plants. After screening 256 AFLP primer combinations 456 pairs SSR primers, we found 14 2 that closely locus. Among these markers, marker CB1022 showed codominant inheritance. By...
The component selection of minimum noise fraction (MNF) rotation transformation is analyzed in terms classification accuracy using support vector machine (SVM) as a classifier for hyper spectral image. Five different group number MNF components are evaluated validation points and map. Further evaluation including error distribution separation-class accuracies comparison performed. experimental result AVIRIS data shows that keep about 1/10 could achieve best accuracies. However, target...