A5107748344- Advanced Graph Neural Networks
- Recommender Systems and Techniques
- Text and Document Classification Technologies
- Cryptography and Data Security
- Complexity and Algorithms in Graphs
- Advanced Text Analysis Techniques
- Topic Modeling
- Chaos-based Image/Signal Encryption
- Sentiment Analysis and Opinion Mining
- Privacy-Preserving Technologies in Data
- Vehicular Ad Hoc Networks (VANETs)
- Domain Adaptation and Few-Shot Learning
- Mobile Ad Hoc Networks
- Image Retrieval and Classification Techniques
- Quantum Information and Cryptography
- Digital Platforms and Economics
- Quantum Computing Algorithms and Architecture
- Spam and Phishing Detection
- Natural Language Processing Techniques
- Cooperative Communication and Network Coding
- Caching and Content Delivery
- Complex Network Analysis Techniques
- Scheduling and Optimization Algorithms
- Epigenetics and DNA Methylation
- Consumer Market Behavior and Pricing
Chinese Academy of Civil Aviation Science and Technology
2024
Japan Aviation Electronics Industry (Japan)
2024
Civil Aviation Administration of China
2023
Huainan Normal University
2022
Nanjing University of Science and Technology
2019-2021
Zhejiang Financial College
2021
Alibaba Group (United States)
2017
Peking University
2009
Graph Embedding methods are aimed at mapping each vertex into a low dimensional vector space, which preserves certain structural relationships among the vertices in original graph. Recently, several works have been proposed to learn embeddings based on sampled paths from graph, e.g., DeepWalk, Line, Node2Vec. However, their only preserve symmetric proximities, could be insufficient many applications, even underlying graph is undirected. Besides, they lack of theoretical analysis what exactly...
With the rapid development of cloud computing and Internet Things (IoT) technology, massive data raises shuttles on network every day. To ensure confidentiality utilization these data, industries com... | Find, read cite all research you need Tech Science Press
Despite that path-based and embedding-based models with knowledge graphs (KGs) achieve better recommendation performance compared other deep learning based methods, such improvement is limited due to a lack of modeling user's dynamic interest. To address this issue, we explore principled model provide semantic understanding each item in historical interest sequence KGs. Specifically, propose multi-granularity method, which on knowledge-enhanced path mining fluctuation signal discovery,...
Hierarchical text classification (HTC) focuses on classifying one into multiple labels, which are organized as a hierarchical taxonomy.Due to its wide involution in realistic scenarios, HTC attracts long-term attention from both industry and academia.However, the high cost of multi-label annotation makes suffer data scarcity problem.In view difficulty balancing controllability structural labels diversity, automatically generating high-quality for is challenging under-explored.To fill this...
Geographical source routing is a promising technique for VANETs, due to its adaptability network dynamics and ability handle topology holes. In traditional geographical algorithms best-known neighbor, typically the neighbor nearest next junction in greedy fashion, designated as hop. This approach may cause two drawbacks: (1) might not receive packet correctly (2) non-neighbor nodes are never given opportunities do forwarding. Due broadcast nature of wireless media, can overhear transmission...
Mingzhe Li, XieXiong Lin, Xiuying Chen, Jinxiong Chang, Qishen Zhang, Feng Wang, Taifeng Zhongyi Liu, Wei Chu, Dongyan Zhao, Rui Yan. Proceedings of the 60th Annual Meeting Association for Computational Linguistics (Volume 1: Long Papers). 2022.
Graph Contrastive Learning (GCL) methods benefit from two key properties: alignment and uniformity, which encourage the representation of related objects together while pushing apart different objects. Most GCL aim to preserve uniformity through random graph augmentation strategies indiscriminately negative sampling. However, their performance is highly sensitive augmentation, requires cumbersome trial-and-error expensive domain-specific knowledge as guidance. Besides, these perform sampling...
In recent years, large language models (LLMs) have driven advances in natural processing. Still, their growing scale has increased the computational burden, necessitating a balance between efficiency and performance. Low-rank compression, promising technique, reduces non-essential parameters by decomposing weight matrices into products of two low-rank matrices. Yet, its application LLMs not been extensively studied. The key to compression lies factorization dimensions allocation. To address...
Dynamic pricing through price promotions has been widely employed by online retailers. We study how a promotion strategy -- offering customers discount for products in their shopping cart affects customer behavior the short and long term on retailing platform. conducted randomized field experiment involving more than 100 million 11,000 retailers with Alibaba Group, world's largest randomly assigned eligible to either receive (treatment group) or not (control group). In term, our program...
Geographical source routing is a promising technique for VANETs, due to its adaptability network dynamics and ability handle topology holes. In traditional geographical algorithms best-known neighbor, typically the neighbor closest next junction in greedy fashion, designated as hop. This approach may cause two drawbacks: (1) might not receive packet correctly (2) non-neighbor nodes are never given opportunities do forwarding. Due broadcast nature of wireless media, can overhear transmission...
The notion personalization lies on the core of a real-world product search system, whose aim is to understand user's intent in fine-grained level. existing solutions mainly achieve this purpose through coarse-grained semantic matching terms query and item's description or collective click correlations. Besides issued query, historical behaviors user would cover lots her personalized interests, which promising avenue alleviate gap between users, items queries. However, as specific domain, are...
Recently, the growth of service platforms brings great convenience to both users and merchants, where search engine plays a vital role in improving user experience by quickly obtaining desirable results via textual queries. Unfortunately, users' uncontrollable customs usually bring vast amounts long-tail queries, which severely threaten capability models. Inspired recently emerging graph neural networks (GNNs) contrastive learning (CL), several efforts have been made alleviating issue...
Grounded on pre-trained language models (PLMs), dense retrieval has been studied extensively plain text. In contrast, there little research retrieving data with multiple aspects using models. the scenarios such as product search, aspect information plays an essential role in relevance matching, e.g., category: Electronics, Computers, and Pet Supplies. A common way of leveraging for multi-aspect is to introduce auxiliary classification objective, i.e., item contents predict annotated value...
Click-through rate (CTR) prediction is a crucial issue in recommendation systems. There has been an emergence of various public CTR datasets. However, existing datasets primarily suffer from the following limitations. Firstly, users generally click different types items multiple scenarios, and modeling scenarios can provide more comprehensive understanding users. Existing only include data for same type single scenario. Secondly, multi-modal features are essential multi-scenario as they...
Contrastive learning has achieved impressive success in generation tasks to militate the "exposure bias" problem and discriminatively exploit different quality of references. Existing works mostly focus on contrastive instance-level without discriminating contribution each word, while keywords are gist text dominant constrained mapping relationships. Hence, this work, we propose a hierarchical mechanism, which can unify hybrid granularities semantic meaning input text. Concretely, first...