A5063856862- Topic Modeling
- Sentiment Analysis and Opinion Mining
- Recommender Systems and Techniques
- Advanced Text Analysis Techniques
- Biomedical Text Mining and Ontologies
- Spam and Phishing Detection
- Text and Document Classification Technologies
- Machine Learning and ELM
- Image Retrieval and Classification Techniques
- Information Retrieval and Search Behavior
- Web Data Mining and Analysis
- Semantic Web and Ontologies
- Advanced Image and Video Retrieval Techniques
- Misinformation and Its Impacts
- Advanced Bandit Algorithms Research
- Microbial Natural Products and Biosynthesis
- Advanced Battery Technologies Research
- Plant tissue culture and regeneration
- Plant biochemistry and biosynthesis
- Multimodal Machine Learning Applications
- Digital Marketing and Social Media
- Fungal Biology and Applications
- Genetically Modified Organisms Research
- Image Processing and 3D Reconstruction
- Advanced Graph Neural Networks
Chinese Academy of Medical Sciences & Peking Union Medical College
2024
Institute of Hematology & Blood Diseases Hospital
2024
Hebei Science and Technology Department
2024
Shanghai Electric (China)
2024
Jingdong (China)
2019-2023
Silicon Valley Community Foundation
2021-2023
NetEase (China)
2021-2022
Beijing Forestry University
2019-2022
State Forestry and Grassland Administration
2022
Northeast Forestry University
2020-2021
Accurate estimation of forest height is crucial for the aboveground biomass and monitoring resources. Remote sensing technology makes it achievable to produce high-resolution maps in large geographical areas. In this study, we produced a 25 m spatial resolution wall-to-wall map Baoding city, north China. We evaluated effects three factors on utilizing four types remote data (Sentinel-1, Sentinel-2, ALOS PALSAR-2, SRTM DEM) with National Forest Resources Continuous Inventory (NFCI) data,...
Video popularity prediction plays a foundational role in many aspects of life, such as recommendation systems and investment consulting. Because its technological economic importance, this problem has been extensively studied for years. However, four constraints have limited most related works' usability. First, feature oriented models are inadequate the social media environment, because videos published with no specific content features, strong cast or famous script. Second, studies assume...
Automatic text summarization for a biomedical concept can help researchers to get the key points of certain topic from large amount literature efficiently. In this paper, we present method generating summary given concept, e.g., H1N1 disease, multiple documents based on semantic relation extraction. Our approach includes three stages: 1) We extract relations in each sentence using knowledge representation tool SemRep. 2) develop relation-level retrieval select most relevant query and...
Due to the rapid growth of network data, authenticity and reliability information have become increasingly important presented challenges. Most methods for fake review detection start with textual features behavioral features. However, they are time-consuming easily detected by fraudulent users. Although most existing neural network-based address problems complex semantics reviews, do not account implicit patterns among users, products; additionally, consider usefulness regarding...
In the business domain,bundling is one of most important marketing strategies to conduct product promotions, which commonly used in online e-commerce and offline retailers. Existing recommender systems mostly focus on recommending individual items that users may be interested in, such as considerable research work collaborative filtering directly models interaction between items. this paper, we target at a practical but less explored recommendation problem named personalized bundle...
Extreme Learning Machine (ELM) has many advantages, such as fast learning speed, good generalization performance and high diagnostic accuracy when it is applied in fault diagnosis, but its classification affected by the two network random parameters-input weights thresholds. Particle swarm optimization (PSO) algorithm characteristics of simple, easy to implement found local optimum quickly. This paper proposes Swarm Optimization optimize parameters obtain electronics system diagnosis based...
Relevance has significant impact on user experience and business profit for e-commerce search platform. In this work, we propose a data-driven framework relevance prediction, by distilling knowledge from BERT related multi-layer Transformer teacher models into simple feed-forward networks with large amount of unlabeled data. The distillation process produces student model that recovers more than 97% test accuracy new queries, at serving cost that's several magnitude lower (latency 150x...
We propose a novel domain-specific generative pre-training (DS-GPT) method for text generation and apply it to the product titleand review summarization problems on E-commerce mobile display.First, we adopt decoder-only transformer architecture, which fitswell fine-tuning tasks by combining input output all to-gether. Second, demonstrate utilizing only small amount of data in related domains is powerful. Pre-training languagemodel from general corpus such as Wikipedia or CommonCrawl requires...
The extreme learning machine (ELM) possesses the advantageous features of fast speed, great generalization performance and high precision. However, randomness parameters will influence its precision greatly. This paper proposes a algorithm which is based on differential evolution (DE-ELM) for parameter optimization ELM. It can optimize two parameters, input weights threshold value, are random-generated in network. experiment selects elliptic filter circuit to build fault model. We extract...
Result relevance scoring is critical to e-commerce search user experience. Traditional information retrieval methods focus on keyword matching and hand-crafted or counting-based numeric features, with limited understanding of item semantic relevance. We describe a highly-scalable feed-forward neural model provide score for (query, item) pairs, using only query title as both click feedback well human ratings labels. Several general enhancements were applied further optimize eval/test metrics,...
In recent years, the biomedical literature has been growing rapidly. These articles provide a large amount of information about proteins, genes and their interactions. Reading such huge is tedious task for researchers to gain knowledge gene. As result, it significant have quick understanding query concept by integrating its relevant resources. gene summary generation, we regard automatic as ranking problem apply method learning rank automatically solve this problem. This paper uses three...
This paper presents a learning to rank based gene summarization method using features such as ontology relevance, topic relevance and TextRank.
Players of online games generate rich behavioral data during gaming. Based on these data, game developers can build a range science applications, such as bot detection and social recommendation, to improve the gaming experience. However, development applications requires cleansing, training sample labeling, feature engineering, model development, which makes use in small medium-sized studios still uncommon. While acquiring supervised learning is costly, unlabeled logs are often continuously...
Immunity reconstitution (IR) is crucial for pediatric patients undergoing hematopoietic stem cell transplantation (HSCT), but the impact of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection on lymphocyte subsets post-transplant remains unclear. Therefore, we assessed immune dynamics in children after SARS-CoV-2 infection.
Nowadays, search engines have become indispensable parts of modern human life, which create hundreds and thousands logs every second throughout the world. With explosive growth online information, a key issue for web service is to better understand user's need through short query match preference as much possible. However, due lack personal information in some scenario huge calculation when seeking relevant user group, personalized becomes quite challenging problem. In this work, we propose...
Graph neural networks (GNNs) have gained great prevalence in tackling various analytical tasks on graph-structured data (i.e., networks). Typical GNNs and their variants adopt a message-passing principle that obtains network representations by the attribute propagates along topology, which however ignores rich textual semantics (e.g., local word-sequence) exist numerous real-world networks. Existing methods for text-rich integrate mainly using internal information such as topics or...
Result relevance prediction is an essential task of e-commerce search engines to boost the utility and ensure smooth user experience. The last few years eyewitnessed a flurry research on use Transformer-style models deep text-match improve relevance. However, these two types ignored inherent bipartite network structures that are ubiquitous in logs, making ineffective. We propose this paper novel Second-order Relevance, which fundamentally different from previous First-order result...