A5100686893- Speech Recognition and Synthesis
- Music and Audio Processing
- Speech and Audio Processing
- Immune Response and Inflammation
- MicroRNA in disease regulation
- Toxoplasma gondii Research Studies
- Cancer-related molecular mechanisms research
- Circular RNAs in diseases
- Natural Language Processing Techniques
- Topic Modeling
- Psoriasis: Treatment and Pathogenesis
- Advanced Text Analysis Techniques
- IL-33, ST2, and ILC Pathways
- Cancer, Hypoxia, and Metabolism
- Text Readability and Simplification
- Renal Diseases and Glomerulopathies
- Advanced biosensing and bioanalysis techniques
- Immune cells in cancer
- Genetic factors in colorectal cancer
- Dysphagia Assessment and Management
- Ferroelectric and Negative Capacitance Devices
- Cell Adhesion Molecules Research
- Inflammation biomarkers and pathways
- Head and Neck Cancer Studies
- Renin-Angiotensin System Studies
University of Hong Kong
2021-2025
Hong Kong Science and Technology Parks Corporation
2025
Alibaba Group (United States)
2023-2025
Alibaba Group (China)
2025
Southwest Medical University
2023
Wuhan University
2023
Renmin Hospital of Wuhan University
2023
City University of Hong Kong, Shenzhen Research Institute
2021-2022
Sun Yat-sen University
2017-2021
First People’s Hospital of Jingmen
2016-2021
Abstract Multiple myeloma (MM) is a prevalent hematologic malignancy characterized by abnormal proliferation of cloned plasma cells. Given the aggressive nature and drug resistance MM cells, identification novel genes could provide valuable insights for treatment. In this study we performed machine learning in RNA microarray data purified cell samples from five independent cohorts with 957 patients, identified O-GlcNAcylation transferase ( OGT ) division cycle 27 CDC27 as key prognostic MM....
Recently, unified speech-text models, such as SpeechGPT, VioLA, and AudioPaLM, have achieved remarkable performance on various speech tasks. These models discretize signals into tokens (speech discretization) use a shared vocabulary for both text tokens. Then they train single decoder-only Transformer mixture of However, these rely the Loss Masking strategy ASR task, which ignores dependency among In this paper, we propose to model in an autoregressive way, similar text. We find that...
Monitoring of key markers for lung cancer detection and tracking acquired drug resistance is critical the management disease. We aim to ascertain value monitoring both total cell free DNA concentrations mutant EGFR content within diverse groups individuals most vulnerable disease.We proposed longitudinal circulating using digital PCR. Circulating present in peripheral blood can be obtained non-invasively provide timely disease status update. 25 heavy smokers 50 patients undergoing TKI...
A novel ICP MS DNA assay based on lanthanide labels and hybridization chain reaction amplification has been developed.
BackgroundMany studies have demonstrated that microRNA-21 (miR-21) acts as an oncogene in the tumourigenesis of a variety tumours and may be involved chemotherapeutic drug resistance tumour cells. In this study, we utilized leukaemia cell line K562 vitro model to investigate whether miR-21 is X-ray irradiation resistance.MethodsRetroviral transduction antisense oligonucleotide transfection were used overexpress or knock down expression, respectively. An MTT assay was measure viability,...
Prior studies diagnose the anisotropy problem in sentence representations from pre-trained language models, e.g., BERT, without fine-tuning. Our analysis reveals that embeddings BERT suffer a bias towards uninformative words, limiting performance semantic textual similarity (STS) tasks. To address this bias, we propose simple and efficient unsupervised approach, Diagonal Attention Pooling (Ditto), which weights words with model-based importance estimations computes weighted average of word...
The well-defined mammalian slp76-signalosome is crucial for T-cell immune response, yet whether exists in invertebrates and how it evolved remain unknown. Here we investigated from an evolutionary perspective amphioxus Branchiostoma belcheri (bb). We proved components bbslp76, bbGADS bbItk are present bbslp76 interacts with bbItk, but differences exist between the interaction manners within of human (h). Specifically, has a unique WW-domain that blocked its association hItk decreased...
The Transformer architecture has significantly advanced deep learning, particularly in natural language processing, by effectively managing long-range dependencies. However, as the demand for understanding complex relationships grows, refining Transformer's becomes critical. This paper introduces Skip-Layer Attention (SLA) to enhance models enabling direct attention between non-adjacent layers. method improves model's ability capture dependencies high-level abstract features and low-level...
In sound event detection (SED), overlapping events pose a significant challenge, as certain can be easily masked by background noise or other events, resulting in poor performance. To address this issue, we propose the text-queried SED (TQ-SED) framework. Specifically, first pre-train language-queried audio source separation (LASS) model to separate tracks corresponding different from input audio. Then, multiple target branches are employed detect individual events. AudioSep is...
Full-duplex spoken dialogue systems significantly advance over traditional turn-based systems, as they allow simultaneous bidirectional communication, closely mirroring human-human interactions. However, achieving low latency and natural interactions in full-duplex remains a significant challenge, especially considering human conversation dynamics such interruptions, backchannels, overlapping speech. In this paper, we introduce novel End-to-End GPT-based model OmniFlatten for conversation,...
In our previous work, we introduced CosyVoice, a multilingual speech synthesis model based on supervised discrete tokens. By employing progressive semantic decoding with two popular generative models, language models (LMs) and Flow Matching, CosyVoice demonstrated high prosody naturalness, content consistency, speaker similarity in in-context learning. Recently, significant progress has been made multi-modal large (LLMs), where the response latency real-time factor of play crucial role...
Amphioxus (e.g., Branchiostoma belcheri, Bb) has recently emerged as a new model for studying the origin and evolution of vertebrate immunity. Mammalian lymphocyte-specific tyrosine kinase (Lck) plays crucial roles in T cell activation, differentiation homeostasis, is reported to phosphorylate both ITIM ITSM PD-1 induce recruitment phosphatases thus inhibitory function PD-1. Here, we identified cloned amphioxus homolog human Lck. By generating using an antibody against BbLck, found that...