A5100733317- Topic Modeling
- Natural Language Processing Techniques
- Domain Adaptation and Few-Shot Learning
- Metaheuristic Optimization Algorithms Research
- Semantic Web and Ontologies
- Neural Networks and Applications
- Machine Learning and ELM
- Artificial Immune Systems Applications
- Advanced Algorithms and Applications
- Biomedical Text Mining and Ontologies
- Evolutionary Algorithms and Applications
- Machine Learning and Data Classification
- Advanced Neural Network Applications
- Multimodal Machine Learning Applications
- Advanced Multi-Objective Optimization Algorithms
- Sentiment Analysis and Opinion Mining
- Speech and Audio Processing
- Image Retrieval and Classification Techniques
- Advanced Text Analysis Techniques
- Cognitive Computing and Networks
- Adversarial Robustness in Machine Learning
- Service-Oriented Architecture and Web Services
- Blind Source Separation Techniques
- Silkworms and Sericulture Research
- Emotion and Mood Recognition
Southwest University
2025
Shanghai University
2018-2025
People's Liberation Army 401 Hospital
2025
Chinese People's Liberation Army
2025
Hefei University of Technology
2020-2024
Xidian University
2020-2024
PLA Academy of Military Science
2023
Academy of Military Medical Sciences
2021
Shanghai University of Engineering Science
2020-2021
National University of Defense Technology
2021
Math word problem (MWP) is challenging due to the limitation in training data where only one “standard” solution available. MWP models often simply fit this rather than truly understand or solve problem. The generalization of (to diverse scenarios) thus limited. To address problem, paper proposes a novel approach, TSN-MD, by leveraging teacher network integrate knowledge equivalent expressions and then regularize learning behavior student network. In addition, we introduce multiple-decoder...
Kernel-based extreme learning machine (KELM), as a natural extension of ELM to kernel learning, has achieved outstanding performance in addressing various regression and classification problems. Compared with the basic ELM, KELM better generalization ability owing no needs number hidden nodes given beforehand random projection mechanism. Since is derived under minimum mean square error (MMSE) criterion for Gaussian assumption noise, its may deteriorate non-Gaussian cases, seriously. To...
Silkworm pupae (SP), the pupal stage of an edible insect, have strong potential in food, medicine, and cosmetic industries. Sex sorting is essential to enhance nutritional content genetic traits SP crossbreeding but it remains labor intensive time consuming. An intelligent method needed urgently improve efficiency productivity. To address problem, automatic sex-separation system was developed based on computer vision deep learning. Specifically, gonad features, a novel real-time sex...
IMPORTANCE The rapid advancement of AI in ophthalmology is transforming traditional diagnostic approaches, especially resource-limited settings. shortage specialized ophthalmologists and lack standardized reporting primary care creates an urgent need for systems capable both automated report generation interactive clinical decision support. OBJECTIVE To develop a multimodal artificial intelligence system (OphthUS-GPT) that integrates BLIP DeepSeek models to enable support from ophthalmic...
Depression has a large impact on one's personal life, especially during the COVID-19 pandemic. People have been trying to develop reliable methods for depression detection task. Recently, based deep learning attracted much attention from research community. However, they still face challenge that data collection and annotation are difficult expensive. In many real-world applications, only small number of or even no training available. this context, we propose Prompt-based Topic-modeling...
Deep learning models often fit undesired dataset bias in training. In this paper, we formulate the using <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">causal inference</i> , which helps us uncover ever-elusive causalities among key factors training, and thus pursue desired causal effect without bias. We start from revisiting process of building a visual recognition system, then propose structural model (SCM) for variables involved...
Zero-shot Visual Question Answering (VQA) is a prominent vision-language task that examines both the visual and textual understanding capability of systems in absence training data. Recently, by converting images into captions, information across multi-modalities bridged Large Language Models (LLMs) can apply their strong zero-shot generalization to unseen questions. To design ideal prompts for solving VQA via LLMs, several studies have explored different strategies select or generate...
This paper describes the information properties of museum specimen labels and machine learning tools to automatically extract Darwin Core (DwC) other metadata from these processed through Optical Character Recognition (OCR). The DwC is a profile describing core set access points for search retrieval natural history collections observation databases. Using HERBIS Learning System (HLS) we 74 independent elements labels. automated text extraction are provided as web service so that users can...
The task of math word problem (MWP) generation, which generates an MWP given equation and relevant topic words, has increasingly attracted researchers’ attention. In this work, we introduce a simple memory retrieval module to search related training MWPs, are used augment the generation. To retrieve more data, also propose disentangled based on module. end, first disentangle MWPs into logical description scenario then record them in respective modules. Later, use words as queries...
In many real-world scenarios, distribution shifts exist in the streaming data across time steps. Many complex sequential can be effectively divided into distinct regimes that exhibit persistent dynamics. Discovering shifted behaviors and evolving patterns underlying are important to understand dynamic system. Existing methods typically train one robust model work for of distributions or sequentially adapt utilizing explicitly given regime boundaries. However, there two challenges: (1)...
Summary A corpus (eg, patents or news texts) is an important knowledge resource that contains various topics, such as specific technologies social events. Topic detection models of corpus, eg, Latent Dirichlet Allocation and KeyGraph, provide basis for exploring the status quo trends in science, technology, However, these suffer from low retrieval performance they only consider text own explicit semantics a single‐domain corpus. In addition, many incremental models, online‐LDA, are based on...
Feature selection based on particle swarm optimization is often employed for promoting the performance of artificial intelligence algorithms. However, its interpretability has been lacking concrete research. Improving stability feature method a way to effectively improve interpretability. A novel approach named Interpretable Particle Swarm Optimization developed in this paper. It uses four data perturbation ways and three filter methods obtain stable subsets, adopts Fuch map convert them...
Abstract High‐dimensional data exists widely in the real world, such as gene, magnetic resonance imaging (MRI), text, web and so on. Feature selection is an effective powerful method that often adopted to reduce dimensions of high‐dimensional for promoting learning algorithm's ability obtain useful information from them. However, feature stability lacks attention a long time, which plays important role getting compelling results. This study proposes stable approach called ant‐antlion...
Feature selection is a widely used data pre-processing method. However, the research on feature stability very rare. Although there are some related studies that mainly focus filters rather than wrappers, especially wrappers based evolutionary algorithms. In this paper, stable method called brain storm optimisation (SBSO) proposed. It puts forward new population initialisation strategy, which treats results as guide information for and utilises such chaos to initialise population. SBSO also...
Silent speech recognition (SSR) is a new application of human-computer interaction based on electromyography (EMG), which solves the limitation acoustic signal dependence. In low signal-to-noise ratio (SNR) environment, traditional methods cannot accurately segment EMG active signal. This paper proposes an energy detection method spectral subtraction backtracking for detecting signals to assist silent recognition. The experiments are mainly detection. addition detection, also used improve...
Feature selection is important for learning algorithms, and it still an open problem. Antlion optimizer excellent nature inspired method, but doesn't work well feature selection. This paper proposes a hybrid approach called Ant-Antlion Optimizer which combines advantages of antlion's smart behavior antlion ant's powerful searching movement ant colony optimization. A mutation operator also adopted to strengthen exploration ability. Comprehensive experiments by binary classification problems...
Multi-objective evolutionary algorithms are widely used in many engineering optimization problems and artificial intelligence applications. Ant lion optimizer is an outstanding method, but two issues need to be solved extend it the multi-objective field, one how update Pareto archive, other choose elite ant lions from archive. We develop a novel variant of this paper. A new measure combining dominance relation distance information individuals put forward tackle first issue. The concept time...
Industrial process data has the characteristics of complexity, variability and noisy, which brings challenges to data-driven production predictive modeling for industrial processes basing on traditional extreme learning machine (ELM). Therefore, this paper proposes an improved ELM based auto-encoder (AE) (AE-ELM). The AE can extract main features with lower-dimension by eliminating linear correlation among original complex data. Then, are used as inputs ELM. For purpose verifying...
Curriculum Learning (CL) selects the training samples from easy to difficult boost classification results. Most existing variates of CL measure difficulty level an example in intuitive way, i.e., loss value between prediction and ground truth. This way ignores different distances every class boundary, which is implied vectors. In this paper, we propose a novel framework, named Balance Loss Learning(BLCL), reveal comprehensive improve curriculum process based on deep architectures. We follow...
Abstract Given a question and its answer candidates (named QA corpus), selection is the task of identifying most relevant answers to question. Answer widely used in answering, web search, so on. Current deep neural network models primarily utilize local features extracted from input question‐answer pairs (QA pairs). However, global contained corpora are under‐utilized, we argue that these substantially contribute task. To verify this point view, propose novel model combines for selection. In...