A5043602228- Cancer Mechanisms and Therapy
- Domain Adaptation and Few-Shot Learning
- Nutrition, Genetics, and Disease
- Machine Learning and Data Classification
- Advanced Neural Network Applications
- Generative Adversarial Networks and Image Synthesis
- Multimodal Machine Learning Applications
- Video Analysis and Summarization
- Occupational and environmental lung diseases
- Interstitial Lung Diseases and Idiopathic Pulmonary Fibrosis
- Topic Modeling
- Advanced Image Processing Techniques
- Inflammation biomarkers and pathways
- RNA modifications and cancer
- Adversarial Robustness in Machine Learning
- Natural Language Processing Techniques
- Digital Media Forensic Detection
- Biomarkers in Disease Mechanisms
- COVID-19 diagnosis using AI
- Circular RNAs in diseases
- Imbalanced Data Classification Techniques
- Fatty Acid Research and Health
- Artificial Intelligence in Games
- Occupational exposure and asthma
- Bone Metabolism and Diseases
West China Medical Center of Sichuan University
2022-2025
Fujian Medical University
2020-2025
The University of Texas at Austin
2021-2024
Sichuan University
2020-2024
China National Center for Food Safety Risk Assessment
2024
University of Science and Technology of China
2020-2021
PLA Navy General Hospital
2013-2015
Transformers have quickly shined in the computer vision world since emergence of Vision (ViTs). The dominant role convolutional neural networks (CNNs) seems to be challenged by increasingly effective transformer-based models. Very recently, a couple advanced models strike back with large kernels motivated local-window attention mechanism, showing appealing performance and efficiency. While one them, i.e. RepLKNet, impressively manages scale kernel size 31x31 improved performance, starts...
Orca 1 learns from rich signals, such as explanation traces, allowing it to outperform conventional instruction-tuned models on benchmarks like BigBench Hard and AGIEval. In 2, we continue exploring how improved training signals can enhance smaller LMs' reasoning abilities. Research small LMs has often relied imitation learning replicate the output of more capable models. We contend that excessive emphasis may restrict potential seek teach employ different solution strategies for tasks,...
Sparsely activated Mixture-of-Experts (MoE) is becoming a promising paradigm for multi-task learning (MTL). Instead of compressing multiple tasks' knowledge into single model, MoE separates the parameter space and only utilizes relevant model pieces given task type its input, which provides stabilized MTL training ultra-efficient inference. However, current approaches adopt fixed network capacity (e.g., two experts in usual) all tasks. It potentially results over-fitting simple tasks or...
With graphs rapidly growing in size and deeper graph neural networks (GNNs) emerging, the training inference of GNNs become increasingly expensive. Existing network weight pruning algorithms cannot address main space computational bottleneck GNNs, caused by connectivity graph. To this end, paper first presents a unified GNN sparsification (UGS) framework that simultaneously prunes adjacency matrix model weights, for effectively accelerating on large-scale graphs. Leveraging new tool, we...
Creating a pro-regenerative immune microenvironment around implant biomaterial surfaces is significant to osseointegration. Immune cells, especially macrophages that participate in the osseointegration, including osteogenesis, osteoclastogenesis, and angiogenesis, should be considered when testing biomaterials. In this study, we immobilized an antimicrobial peptide GL13K with immunomodulatory properties onto titanium surface via silanization. The modified show good biocompatibility bone...
Alignment of large language models (LLMs) with human values and preferences, often achieved through fine-tuning based on feedback, is essential for ensuring safe responsible AI behaviors. However, the process typically requires substantial data computation resources. Recent studies have revealed that alignment might be attainable at lower costs simpler methods, such as in-context learning. This leads to question: Is predominantly superficial? In this paper, we delve into question provide a...
Background Periodontitis is a persistent inflammatory condition that mainly affects the supportive tissues of teeth, comprising gingiva, periodontal ligament, and alveolar bone. Purpose The present study focused on addressing fisetin’s therapeutic activities against experimental periodontitis model in rats. Materials Methods Experimental was initiated rats by ligation procedure. fisetin treatment given with three different dosages for 10 days, finally, were sacrificed 11th day, subsequently,...
Many real-world applications have to tackle the Positive-Unlabeled (PU) learning problem, i.e., binary classifiers from a large amount of unlabeled data and few labeled positive examples. While current state-of-the-art methods employ importance reweighting design various risk estimators, they ignored capability model itself, which could provided reliable supervision. This motivates us propose novel Self-PU framework, seamlessly integrates PU self-training. highlights three "self"-oriented...
The polarization of macrophages and its anti-inflammatory proinflammatory properties play a significant role in host response after implant placement to determine the outcome osseointegration long-term survival. In previous study, we immobilized an antimicrobial peptide, GL13K, onto titanium surfaces provide immune regulation property. herein presented aimed at investigating whether GL13K surface could improve osteogenesis reduce inflammatory reaction around biomaterials by altering...
There have been long-standing controversies and inconsistencies over the experiment setup criteria for identifying "winning ticket" in literature. To reconcile such, we revisit definition of lottery ticket hypothesis, with comprehensive more rigorous conditions. Under our new definition, show concrete evidence to clarify whether winning exists across major DNN architectures and/or applications. Through extensive experiments, perform quantitative analysis on correlations between tickets...
The plant-based medicinal food (PBMF) is a functional compound extracted from 6 and edible plants: Coix seed, L. edodes, A. officinalis L., H. cordata, Dandelion, G. frondosa. Our previous studies have confirmed that the PBMF possesses anti-tumor properties in subcutaneous xenograft model of nude mice. This study aims to further investigate effects potential molecular mechanisms on recurrence metastasis gastric cancer (GC).Postoperative GC was successfully established inbred 615 mice...
PM2.5 is a well-known air pollutant threatening public health. Studies confirmed that exposure to the particles could impair pulmonary function, cause chronic obstructive disease, and increase incidence of lung cancer. The characteristic varies across regions. toxic function in southwest China remains be elucidated. This study aimed investigate injury its mechanisms induced by collected Chengdu. Rats were administered with intratracheal instillation for 4 weeks. Biochemical, cell count,...
Natural products, especially those with high contents of phytochemicals, are promising alternative medicines owing to their antitumor properties and few side effects. In this study, the effects a plant-based medicinal food (PBMF) composed six edible plants, namely, Coix seed, Lentinula edodes, Asparagus officinalis L., Houttuynia cordata, Dandelion, Grifola frondosa, on gastric cancer underlying molecular mechanisms were investigated in vivo.A subcutaneous xenograft model was successfully...
Deep generative adversarial networks (GANs) have gained growing popularity in numerous scenarios, while usually suffer from high parameter complexities for resource-constrained real-world applications. However, the compression of GANs has less been explored. A few works show that heuristically applying techniques normally leads to unsatisfactory results, due notorious training instability GANs. In parallel, lottery ticket hypothesis shows prevailing success on discriminative models, locating...
Nicotinamide mononucleotide (NMN) is a natural antioxidant approved as nutritional supplement and food ingredient, but its protective role in silicosis characterized by oxidative damage remains unknown. In this study, we generated model intratracheal instillation of silica, then performed histopathological, biochemical, transcriptomic analysis to evaluate the NMN silicosis. We found that mitigated lung at 7 28 days, manifested decreasing coefficient weight histological changes, alleviated...
In the rapidly advancing arena of large language models (LLMs), a key challenge is to enhance their capabilities amid looming shortage high-quality training data. Our study starts from an empirical strategy for light continual LLMs using original pre-training data sets, with specific focus on selective retention samples that incur moderately high losses. These are deemed informative and beneficial model refinement, contrasting highest-loss samples, which would be discarded due correlation...