A5058154698- Advanced Graph Neural Networks
- Topic Modeling
- Recommender Systems and Techniques
- Limits and Structures in Graph Theory
- Advanced Text Analysis Techniques
- Sentiment Analysis and Opinion Mining
- Advanced Graph Theory Research
- Complex Network Analysis Techniques
- Text and Document Classification Technologies
- Software Reliability and Analysis Research
- Advanced Memory and Neural Computing
- Graph theory and applications
- Fault Detection and Control Systems
- Natural Language Processing Techniques
- Advanced Image and Video Retrieval Techniques
- Coding theory and cryptography
- Liver Disease Diagnosis and Treatment
- Physical Unclonable Functions (PUFs) and Hardware Security
- Advanced Clustering Algorithms Research
- Ferroelectric and Negative Capacitance Devices
- semigroups and automata theory
- Drug-Induced Hepatotoxicity and Protection
- Image Retrieval and Classification Techniques
- Image Processing Techniques and Applications
- Analytic Number Theory Research
Microsoft Research Asia (China)
2025
Beijing University of Posts and Telecommunications
2018-2024
National University of Singapore
2023
North China Electric Power University
2022
Meizu (China)
2022
Hu Linmei, Tianchi Yang, Chuan Shi, Houye Ji, Xiaoli Li. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint (EMNLP-IJCNLP). 2019.
Short text classification has been widely explored in news tagging to provide more efficient search strategies and effective results for information retrieval. However, most existing studies, concentrating on long classification, deliver unsatisfactory performance short texts due the sparsity issue insufficiency of labeled data. In this article, we propose a novel heterogeneous graph neural network-based method semi-supervised leveraging full advantage limited data large unlabeled through...
Linmei Hu, Tianchi Yang, Luhao Zhang, Wanjun Zhong, Duyu Tang, Chuan Shi, Nan Duan, Ming Zhou. Proceedings of the 59th Annual Meeting Association for Computational Linguistics and 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.
Heterogeneous graphs (HGs), consisting of multiple types nodes and links, can characterize a variety real-world complex systems. Recently, heterogeneous graph neural networks (HGNNs), as powerful embedding method to aggregate structure attribute information, has earned lot attention. Despite the ability HGNNs in capturing rich semantics which reveal different aspects nodes, they still stay at coarse-grained level simply exploits structural characteristics. In fact, unstructured text content...
Since heterogeneous information network (HIN) is able to integrate complex and contain rich semantics, there a surge of HIN based recommendation in recent years. Although existing methods have achieved performance improvement some extent, they still face the following problems: how extensively exploit comprehensively explore local global for recommendation. To address these issues, we propose unified model LGRec fuse top-N HIN. We firstly most informative neighbor users items respectively...
With the rapid growth of interaction data, many clustering methods have been proposed to discover patterns as prior knowledge beneficial downstream tasks. Considering that an can be seen action occurring among multiple objects, most existing model objects and their pair-wise relations nodes links in graphs. However, they only leverage part information real entire interactions, i.e., either decompose into several sub-interactions for simplification, or focus on some specific types which...
With the great popularity of Graph Neural Networks (GNNs), their robustness to adversarial topology attacks has received significant attention. Although many attack methods have been proposed, they mainly focus on fixed-budget attacks, aiming at finding most perturbations within a fixed budget for target node. However, considering varied each node, there is an inevitable dilemma caused by budget, i.e., no successful perturbation found when relatively small, while if it too large, yielding...
A graph is called $k$-critical if its chromatic number $k$ but any proper subgraph has less than $k$. An old and important problem in theory asks to determine the maximum of edges an $n$-vertex graph. This widely open for integer $k\geq 4$. Using a structural characterization Greenwell Lov\'asz extremal result Simonovits, Stiebitz proved 1987 that 4$ sufficiently large $n$, this balanced complete $(k-2)$-partite In paper we obtain first improvement on above past 35 years. Our proofs combine...
<sec> <title>BACKGROUND</title> Drug-induced liver injury (DILI) is associated with treatment discontinuation and failure in patients tuberculosis (TB). Due to the unpredictable nature of DILI, it increasingly becoming a threat global TB eradication initiatives. </sec> <title>OBJECTIVE</title> This study aimed develop validate an interpretable prediction model for DILI during treatment. <title>METHODS</title> Eligible adult Ningbo City from 2015 2020 were included. We used XGBoost algorithm...
Abstract A graph is called $k$ -critical if its chromatic number but every proper subgraph has less than . An old and important problem in theory asks to determine the maximum of edges an $n$ -vertex graph. This widely open for integer $k\geq 4$ Using a structural characterisation Greenwell Lovász extremal result Simonovits, Stiebitz proved 1987 that sufficiently large , this balanced complete $(k-2)$ -partite In paper, we obtain first improvement above past 35 years. Our proofs combine...
Analysis over product reviews has drawn much attention due to its wide application. Most of the sentiment analysis research focuses on entertainment and catering limitation existing public datasets. In order promote comprehensiveness data in field analysis, we present a new large-scale multi-sentiment tobacco dataset by distilling effective consumer experience information from massive online consumption. The release this would push forward field. With goal advancing facilitating overall...
We obtain some new upper bounds on the maximum number $f(n)$ of edges in $n$-vertex graphs without containing cycles length four. This leads to an asymptotically optimal bound for a broad range integers $n$ as well disproof conjecture Erd\H{o}s from 1970s which asserts that $f(n)=\frac12 n^{3/2}+\frac14 n+o(n)$.