A5101746453- Blockchain Technology Applications and Security
- Optical measurement and interference techniques
- Optical Systems and Laser Technology
- Crime, Illicit Activities, and Governance
- Adaptive optics and wavefront sensing
- CAR-T cell therapy research
- Natural Language Processing Techniques
- Topic Modeling
- Cell Image Analysis Techniques
- Advanced Graph Neural Networks
- AI in cancer detection
- Advanced X-ray Imaging Techniques
- Geographic Information Systems Studies
- Data Management and Algorithms
- Digital Platforms and Economics
- Imbalanced Data Classification Techniques
- Caching and Content Delivery
- Cybercrime and Law Enforcement Studies
- Spam and Phishing Detection
- Advanced Fiber Optic Sensors
- Software-Defined Networks and 5G
- Cybersecurity and Information Systems
- Semantic Web and Ontologies
- Public Relations and Crisis Communication
- Diverse Musicological Studies
Zhejiang University
2023-2024
Singapore Management University
2019-2024
Korea University
2023-2024
Xiaomi (China)
2023
National Taiwan University of Science and Technology
2021-2023
Beijing Institute of Graphic Communication
2022
Soochow University
2017-2021
First Affiliated Hospital of Soochow University
2021
Beijing University of Posts and Telecommunications
2018-2020
National Chengchi University
2018-2019
A new algorithm based on deep learning analyzes angiogenic morphogenesis images taken from angiogenesis a chip. This method can assess the morphology of in great depth using multiple indicators and extract 3D indices 2D images.
Existing research on federated learning has been focused the setting where is coordinated by a centralized entity. Yet greatest potential of future collaborative intelligence would be unleashed in more open and democratized with no central entity dominant role, referred to as "decentralized learning". New challenges arise accordingly achieving both correct model training fair reward allocation collective effort among all participating nodes, especially threat Byzantine node jeopardising...
Summary In this paper, we propose an effective model for the similarity metrics of English sentences. model, first make use word embedding and convolutional neural network (CNN) to produce a sentence vector then leverage information pair calculate score similarity. Considering case long‐range semantic dependencies between words, novel method transforming embeddings construct three‐dimensional feature tensor. addition, incorporate k‐max pooling into adapt variable lengths input The proposed...
Nowadays, users of social networks like tweets and weibo have generated massive geo-tagged records, these records reveal their activities in the physical world together with spatio-temporal dynamics. Existing trajectory data management studies mainly focus on analyzing properties trajectories, while leaving understanding largely untouched. In this paper, we incorporate semantic analysis activity information embedded trajectories into query modelling processing, aim providing end more...
In traditional networks, routing table is essential for packet transmission due to the lack of direction information about destination in head packet. However, it feasible make address device encode with application data technology. this article, we propose new identities networking routers -vectors, and a principle based on these vectors designed accordingly. These distance serve as pattern network topology. Then, decisions could be made by vector calculations only requirement query...
Knowledge base question answering (KBQA) is a challenging task that aims to retrieve correct answers from large-scale knowledge bases. Existing attempts primarily focus on entity representation and final answer reasoning, which results in limited supervision for this task. Moreover, the relations, empirically determine reasoning path selection, are not fully considered recent advancements. In study, we propose novel framework, RE-KBQA, utilizes relations enhance introduce additional...
Organoids and 3D imaging techniques are crucial for studying human tissue structure function, but traditional reconstruction methods expensive time consuming, relying on complete z stack confocal microscopy data. This paper introduces VONet, a deep learning-based system organoid rendering that uses fully convolutional neural network to reconstruct entire structures from minimal number of images. VONet was trained library over 39,000 virtual organoids (VOs) with diverse structural features...
For dentists, it is very important to determine the color of denture. Shade selection in dental practice an and difficult task. In shade matching process, will be affected by observer's physiological conditions such as age, mood, fatigue, so on. These make a difference on judgement between actual teeth color. past, dentists use tabs reference basis match intra-oral environment. this paper, efficient analysis methodology based image processing fuzzy decision techniques proposed for matching....
To explore the immune cell therapy for T lymphoma, we developed CD4-specific chimeric antigen receptor- (CAR-) engineered cells (CD4CART), and cytotoxic effects of CD4CART were determined in vitro vivo.
Data is one of the most critical resources in AI Era. While substantial research has been dedicated to training machine learning models using various types data, much less efforts have invested exploration assessing and governing data assets end-to-end processes science, that is, pipeline where collected processed, then are produced, requested, deployed, shared evolved. To provide a state-of-the-art overall picture this important novel area advocate related development, we present tutorial...
With the boom of cryptocurrency and its concomitant financial risk concerns, detecting fraudulent behaviors associated malicious addresses has been drawing significant research effort. Most existing studies, however, rely on full history features or full-fledged address transaction networks, both which are unavailable in problem early detection therefore failing them for task. To detect stage, we present Evolve Path Tracer, consists Encoder LSTM, Graph GCN, Hierarchical Survival Predictor....
Ronchi lateral shearing interferometry is a promising wavefront sensing technology with the advantages of simple structure and no reference light, which can realize high-precision aberration measurement. To obtain shear information in both directions, conventional double-Ronchi interferometer sequentially applies two orthogonal one-dimensional gratings as object-plane splitting element optics under test. Simultaneously, another grating positioned on image plane same orientation to capture...
Abstract The KRASG12D mutation is an ideal target for anti-cancer therapies as its expression typically clonal, restricted to cancer tissue, and among the most common oncogenic drivers in solid tumors. TCR-T cell have demonstrated clinical activity some cancers but been limited by heterogeneous antigen unfavorable tumor microenvironments. By targeting which has established genetic dependency, AFNT-212 designed selectively all cells while avoiding on-target/off-tumor toxicities. non-virally...
Federated recommender systems (FedRSs) effectively tackle the trade-off between recommendation accuracy and privacy preservation. However, recent studies have revealed severe vulnerabilities in FedRSs, particularly against untargeted attacks seeking to undermine their overall performance. Defense methods employed traditional are not applicable existing robust aggregation schemes for other federated learning-based applications proven ineffective FedRSs. Building on observation that malicious...
Nowadays, more and researchers are paying their attention to green routing. In this paper, we consider power consumption as a kind of QoS (quality service) apply new learning-based approach for energy efficient transportation Compared with traditional rule-based methods, the proposed method can learn additional information from networks improve routing performance, have flexibility meet different requirements. First, propose identification network nodes, namely node vectors, basic algorithm...
Cryptocurrency has been subject to illicit activities probably more often than traditional financial assets due the pseudo-anonymous nature of its transacting entities. An ideal detection model is expected achieve all three critical properties early detection, good interpretability, and versatility for various activities. However, existing solutions cannot meet these requirements, as most them heavily rely on deep learning without interpretability are only available retrospective analysis a...