A5101663600- Advanced Neural Network Applications
- Tuberculosis Research and Epidemiology
- Methane Hydrates and Related Phenomena
- Genomics and Phylogenetic Studies
- interferon and immune responses
- Bacteriophages and microbial interactions
- Mycobacterium research and diagnosis
- Adversarial Robustness in Machine Learning
- Solar-Powered Water Purification Methods
- Membrane Separation Technologies
- Microbial Community Ecology and Physiology
- Topic Modeling
- Cytomegalovirus and herpesvirus research
- Advanced Memory and Neural Computing
- Aerogels and thermal insulation
- IoT and Edge/Fog Computing
- Spacecraft and Cryogenic Technologies
- Hydrocarbon exploration and reservoir analysis
- Explainable Artificial Intelligence (XAI)
- Anomaly Detection Techniques and Applications
- NF-κB Signaling Pathways
- Advanced Text Analysis Techniques
- Single-cell and spatial transcriptomics
- Thermal Regulation in Medicine
- Cancer Genomics and Diagnostics
Harbin Institute of Technology
2021-2024
Tongji University
2018-2024
University of Southern California
2022-2024
WuXi AppTec (China)
2021
Shanghai Pulmonary Hospital
2018-2019
Tsinghua University
2015
South China University of Technology
2015
Adaptation to hypoxia is a major challenge for the survival of Mycobacterium tuberculosis (Mtb) in vivo. Interferon (IFN)-γ-producing CD8
Tuberculosis caused by Mycobacterium tuberculosis (Mtb) infection remains a large global public health problem. One striking characteristic of Mtb is its ability to adapt hypoxia and trigger the ensuing transition dormant state for persistent infection, but how response regulated largely unknown. Here we performed quantitative acetylome analysis compare acetylation profile under aeration hypoxia, showed that 377 sites in 269 proteins were significantly changed hypoxia. In particular,...
Abstract Pathogenic mycobacteria induce the formation of hypoxic granulomas during latent tuberculosis (TB) infection, in which immune system contains, but fails to eliminate mycobacteria. Fatty acid metabolism-related genes are relatively overrepresented mycobacterial genome and favor host-derived fatty acids as nutrient sources. However, whether how modulate host metabolism drive granuloma progression remains unknown. Here, we report that under hypoxia markedly secrete protein...
Abstract Sequence classification facilitates a fundamental understanding of the structure microbial communities. Binary metagenomic sequence classifiers are insufficient because environmental metagenomes typically derived from multiple sources. Here we introduce deep-learning based classifier, DeepMicroClass, that classifies contigs into five classes, i.e. viruses infecting prokaryotic or eukaryotic hosts, chromosomes, and plasmids. DeepMicroClass achieved high performance for all classes at...
"Rebooting Computing" (RC) is an effort in the IEEE to rethink future computers. RC started 2012 by co-chairs, Elie Track (IEEE Council on Superconductivity) and Tom Conte (Computer Society). takes a holistic approach, considering revolutionary as well evolutionary solutions needed advance computer technologies. Three summits have been held 2013 2014, discussing different technologies, from emerging devices user interface, security energy efficiency, neuromorphic reversible computing. The...
“Rebooting Computing” (RC) is an effort in the IEEE to rethink future computers. RC started 2012 by co-chairs, Elie Track (IEEE Council on Superconductivity) and Tom Conte (Computer Society). takes a holistic approach, considering revolutionary as well evolutionary solutions needed advance computer technologies. Three summits have been held 2013 2014, discussing different technologies, from emerging devices user interface, security energy efficiency, neuromorphic reversible computing. The...
Abstract Sequence classification reduces the complexity of metagenomes and facilitates a fundamental understanding structure function microbial communities. Binary metagenomic classifiers offer an insufficient solution because environmental are typically derived from multiple sequence sources, including prokaryotes, eukaryotes viruses both. Here we introduce deep-learning based (as opposed to alignment-based) classifier, DeepMicroClass, that classifies contigs into five classes, i.e.,...
Phage-host associations play important roles in microbial communities. But natural communities, as opposed to culture-based lab studies where phages are discovered and characterized metagenomically, their hosts generally not known. Several programs have been developed for predicting which phage infects host based on various sequence similarity measures or machine learning approaches. These often whole viral genomes, but metagenomics-based studies, we rarely genomes rather must rely contigs...
The nature of comments usually has an important impact on the network environment. Polite and gentle can not only promote communication between users, but also maintain stability platform. On contrary, rude toxic will make environment unacceptable. Therefore, we need to impose certain restrictions comments. This article is based XLM-RoBERTa model achieve classification multilingual We first use training verification data train optimize model, then test get final results. In addition, our...
Poisoning attack in which an adversary misleads the learning process by manipulating its training set significantly affect performance of classifiers security applications. This paper proposed a robust method reduces influences samples on learning. The sensitivity, defined as fluctuation output with small perturbation input, Localized Generalization Error Model (L-GEM) is measured for each sample. classifier's may be sensitive and inaccurate since these are different from other untainted...
Transformer verification draws increasing attention in machine learning research and industry. It formally verifies the robustness of transformers against adversarial attacks such as exchanging words a sentence with synonyms. However, performance transformer is still not satisfactory due to bound-centric computation which significantly different from standard neural networks. In this paper, we propose Faith, an efficient framework for on GPUs. We first semantic-aware graph transformation...
Text is the main carrier of information on Internet. More and more researchers are committed to mining textual for great value. In terms education, text can greatly help teachers students improve their writing. Computers assist in grading classifying students' argumentative essays into three kinds: effective, adequate, ineffective. this paper, we use data consisting discourse annotation provided by Kaggle platform. We utilize DeBERTa Predicting Effective Arguments. represent related work...
Abstract Background: Tumors are complex ecosystems composed of different cell types with phenotypes, status and gene profiles. Commonly used GEP tools like bulk RNA Sequencing could only display the expression profiles as a whole, cannot reflect heterogeneous tumor change or immune composition in tumor. Single can be good tool to implement it single level. However, method experiment system should optimized make sure result is reliable interpretable. Thus we try develop optimize our scRNAseq...