A5100705676- Face and Expression Recognition
- Advanced Image and Video Retrieval Techniques
- Advanced Algorithms and Applications
- Radiomics and Machine Learning in Medical Imaging
- Domain Adaptation and Few-Shot Learning
- Advanced Vision and Imaging
- Image Retrieval and Classification Techniques
- Video Surveillance and Tracking Methods
- Neural Networks and Applications
- Blind Source Separation Techniques
- Sparse and Compressive Sensing Techniques
- Remote-Sensing Image Classification
- Text and Document Classification Technologies
- Robotics and Sensor-Based Localization
- Human Pose and Action Recognition
- Topic Modeling
- Remote Sensing and Land Use
- Advanced Computational Techniques and Applications
- Glioma Diagnosis and Treatment
- Natural Language Processing Techniques
- Anomaly Detection Techniques and Applications
- Hand Gesture Recognition Systems
- Speech and Audio Processing
- Software Engineering Research
- Speech Recognition and Synthesis
Dalian University of Technology
2015-2024
Beijing University of Posts and Telecommunications
2006-2024
University of Pennsylvania
2021-2024
California University of Pennsylvania
2024
State Grid Corporation of China (China)
2022-2023
Huazhong University of Science and Technology
2023
Philadelphia University
2023
Northeastern University
2023
Chongqing Institute of Green and Intelligent Technology
2023
National Institute for Viral Disease Control and Prevention
2020-2022
The sudden outbreak of novel coronavirus 2019 (COVID-19) increased the diagnostic burden radiologists. In time an epidemic crisis, we hoped artificial intelligence (AI) to help reduce physician workload in regions with outbreak, and improve diagnosis accuracy for physicians before they could acquire enough experience new disease. Here, present our building deploying AI system that automatically analyzes CT images detect COVID-19 pneumonia features. Different from conventional medical AI,...
Clustering on multi-view data has attracted much more attention in the past decades. Most previous studies assume that each instance appears all views, or there is at least one view containing instances. However, real world often suffers from missing some instances view, leading to research problem of partial clustering. To address this issue, paper proposes a simple yet effective Anchorbased Partial Multi-view (APMC) method, which utilizes anchors reconstruct instance-to-instance...
In the past decade, various sparse learning based unsupervised feature selection methods have been developed. However, most existing studies adopt a two-step strategy, i.e., selecting top-m features according to calculated descending order and then performing K-means clustering, resulting in group of sub-optimal features. To address this problem, we propose Dependence Guided Unsupervised Feature Selection (DGUFS) method select partition data joint manner. Our proposed enhances...
Dictionary learning (DL) has been successfully applied to various pattern classification tasks in recent years. However, analysis dictionary (ADL), as a major branch of DL, not yet fully exploited due its poor discriminability. This paper presents novel DL method, namely Discriminative Analysis Learning (DADL), improve the performance ADL. First, code consistent term is integrated into basic model Second, triplet constraint-based local topology preserving loss function introduced capture...
In this paper, we study oracle character recognition and general sketch recognition. First, a data set of characters, which are the oldest hieroglyphs in China yet remain part modern Chinese is collected for analysis. Second, typical visual representations shape- sketch-related works evaluated. We analyze problems suffered when addressing these determine several representation design criteria. Based on analysis, propose novel hierarchical that combines Gabor-related low-level...
Microlending can lead to improved access capital in impoverished countries. Recommender systems could be used microlending provide efficient and personalized service lenders. However, increasing concerns about discrimination machine learning hinder the application of recommender microfinance industry. Most previous focus on pure personalization, with fairness issue largely ignored. A desirable property is give borrowers from different demographic groups a fair chance being recommended, as...
Unsupervised feature selection has shown significant potential in distance-based clustering tasks. This paper proposes a novel triplet induced method. Firstly, triplet-based loss function is introduced to enforce the selected groups preserve ordinal locality of original data, which contributes Secondly, we simplify orthogonal basis by imposing an constraint on projection matrix. Consequently, general framework for simultaneous and discussed. Thirdly, alternating minimization algorithm...
Abstract A recently developed pneumonia caused by SARS-CoV-2 has quickly spread across the world. Unfortunately, a simplified risk score that could easily be used in primary care or general practice settings not been developed. The objective of this study is to identify triage severe COVID-19 patients. All and critical adult patients with laboratory-confirmed on West campus Union Hospital, Wuhan, China, from 28 January 2020 29 February were included study. Clinical data laboratory results...
Machine learning (ML) integrated with medical imaging has introduced new perspectives in precision diagnostics of high-grade gliomas, through radiomics and radiogenomics. This raised hopes for characterizing noninvasive vivo biomarkers prediction patient survival, tumor recurrence, genomics therefore encouraging treatments tailored to individualized needs. Characterization infiltration based on pre-operative multi-parametric magnetic resonance (MP-MRI) scans may allow the loci future...
Abstract Glioblastoma is a highly heterogeneous disease, with variations observed at both phenotypical and molecular levels. Personalized therapies would be facilitated by non-invasive in vivo approaches for characterizing this heterogeneity. In study, we developed unsupervised joint machine learning between radiomic genomic data, thereby identifying distinct glioblastoma subtypes. A retrospective cohort of 571 IDH-wildtype patients were included the pre-operative multi-parametric MRI scans...
Few-shot classification of remote sensing images has attracted attention due to its important applications in various fields. The major challenge few-shot image scene is that limited labeled samples can be utilized for training. This may lead the deviation prototype feature expression, and thus performance will impacted. To solve these issues, a calibration with feature-generating model proposed classification. In framework, encoder self-attention developed reduce influence irrelevant...
In this paper, we propose a novel Human-like Semantic Cognition Network (HSCN) for aspect-level sentiment classification, motivated by the principles of human beings’ reading cognitive process (pre-reading, active reading, post-reading). We first design word-level interactive perception module to capture correlation between context words and given target words, which can be regarded as pre-reading. Second, mimic targetaware semantic distillation produce targetspecific representation...
In the past decade, various unsupervised hashing methods have been developed for cross-modal retrieval. However, in real-world applications, it is often incomplete case that every modality of data may suffer from some missing samples. Most existing works assume object appears both modalities, hence they not work well partial multi-modal data. To address this problem, we propose a novel Collective Affinity Learning Method (CALM), which collectively and adaptively learns an anchor graph...
Global convergence of the sequential minimal optimization (SMO) algorithm for support vector regression (SVR) is studied in this paper. Given l training samples, SVR formulated as a convex quadratic programming (QP) problem with pairs variables. We prove that if two variables violating optimality condition are chosen update each step and subproblems solved certain way, then SMO always stops within finite number iterations after finding an optimal solution. Also, efficient implementation...
In this paper, a novel unsupervised method for learning sparse features combined with support vector machines classification is proposed. The classical SVM has restrictions on the large-scale applications. This model uses auto encoder, deep algorithm, to improve performance. Firstly, we use multiple layers of encoder learn data. Secondly, classify. Many experimental results show that compared SVM, our proposed can rate. particular, it effectively deal data sets.
Background There have been no researches assessing the research trends of application artificial intelligence in glioma with bibliometric methods. Purpose The aim study is to assess analysis. Methods Documents were retrieved from web science between 1996 and 2022. bibliometrix package Rstudio was applied for data analysis plotting. Results A total 1081 documents annual growth rate 30.47%. top 5 most productive countries USA, China, Germany, France, UK. USA China strongest international...
Dictionary learning (DL) plays an important role in pattern classification. However, a discriminative dictionary has not been well addressed analysis (ADL). This letter proposes Class-aware Analysis Learning (CADL) model to improve the classification performance of conventional ADL. The objective function CADL mainly includes two parts promote discriminability. first part aims learn subdictionary for each class instead global all classes. learned is class-aware, generating block-diagonal...
In the past decade, various multi-view outlier detection methods have been designed to detect horizontal outliers that exhibit inconsistent across-view characteristics. The existing works assume all objects are present in views. However, real-world applications, it is often incomplete case every view may suffer from some missing samples, resulting partial difficult from. To address this problem, we propose a novel Collective Learning (CL) based framework data self-guided way. More...
Semantic image segmentation can be accomplished by assigning a proper object category label to each meaningful region of an image. Beyond the original bottom-up models, use top-down categorization information has been applied semantic improve performance. An excellent example such scheme is integrate Conditional Random Field (CRF) model with sparse dictionary learning. However, existing solutions merely consider discrimination dictionaries obtain better codes, without considering inherent...
Sketch is used for rendering the visual world since prehistoric times, and has become ubiquitous nowadays with increasing availability of touchscreens on portable devices. However, how to automatically map images sketches, a problem that profound implications applications such as sketch-based image retrieval, still remains open. In this paper, we propose novel method draws sketch from single natural image. extraction posed within an unified contour grouping framework, where perceptual first...
Bezigons, i.e., closed paths composed of Bézier curves, have been widely employed to describe shapes in image vectorization results. However, most existing techniques infer the bezigons by simply approximating an intermediate vector representation (such as polygons). Consequently, resultant are sometimes imperfect due accumulated errors, fitting ambiguities, and a lack curve priors, especially for low-resolution images. In this paper, we novel method vectorizing clipart contrast previous...