A5101783515- Speech Recognition and Synthesis
- Speech and Audio Processing
- Cholangiocarcinoma and Gallbladder Cancer Studies
- Image and Signal Denoising Methods
- Natural Language Processing Techniques
- Advanced Neural Network Applications
- Radiomics and Machine Learning in Medical Imaging
- AI in cancer detection
- Tracheal and airway disorders
- Hepatocellular Carcinoma Treatment and Prognosis
- Cell Image Analysis Techniques
- Music and Audio Processing
- Cancer Diagnosis and Treatment
- Topic Modeling
- Neural Networks and Applications
- Advanced Image Processing Techniques
- Stochastic Gradient Optimization Techniques
- Advanced Image and Video Retrieval Techniques
- Bayesian Modeling and Causal Inference
- Pancreatic and Hepatic Oncology Research
- Handwritten Text Recognition Techniques
- Artificial Intelligence in Healthcare
- Cell death mechanisms and regulation
- Advanced Image Fusion Techniques
- Gallbladder and Bile Duct Disorders
Sun Yat-sen University
2023-2025
Jilin University
2017-2025
First Hospital of Jilin University
2017-2025
Beijing Tsinghua Chang Gung Hospital
2020-2024
Tsinghua University
2007-2024
Shanghai University of Traditional Chinese Medicine
2024
State Administration of Traditional Chinese Medicine of the People's Republic of China
2024
Jishou University
2024
Hunan Vocational Institute of Technology
2024
Second Xiangya Hospital of Central South University
2022-2023
A text-to-speech synthesis system typically consists of multiple stages, such as a text analysis frontend, an acoustic model and audio module.Building these components often requires extensive domain expertise may contain brittle design choices.In this paper, we present Tacotron, end-to-end generative that synthesizes speech directly from characters.Given <text, audio> pairs, the can be trained completely scratch with random initialization.We several key techniques to make...
In this work, we propose "global style tokens" (GSTs), a bank of embeddings that are jointly trained within Tacotron, state-of-the-art end-to-end speech synthesis system. The with no explicit labels, yet learn to model large range acoustic expressiveness. GSTs lead rich set significant results. soft interpretable "labels" they generate can be used control in novel ways, such as varying speed and speaking - independently the text content. They also for transfer, replicating single audio clip...
We propose an end-to-end trainable network that can simultaneously detect and recognize text of arbitrary shape, making substantial progress on the open problem reading scene irregular shape. formulate shape detection as instance segmentation problem; attention model is then used to decode textual content each irregularly shaped region without rectification. To extract useful features from image scale features, we a simple yet effective RoI masking step. Additionally, show predictions...
A text-to-speech synthesis system typically consists of multiple stages, such as a text analysis frontend, an acoustic model and audio module. Building these components often requires extensive domain expertise may contain brittle design choices. In this paper, we present Tacotron, end-to-end generative that synthesizes speech directly from characters. Given <text, audio> pairs, the can be trained completely scratch with random initialization. We several key techniques to make...
To understand the dynamics of optimization in deep neural networks, we develop a tool to study evolution entire Hessian spectrum throughout process. Using this, number hypotheses concerning smoothness, curvature, and sharpness learning literature. We then thoroughly analyze crucial structural feature spectra: non-batch normalized observe rapid appearance large isolated eigenvalues spectrum, along with surprising concentration gradient corresponding eigenspaces. In batch these two effects are...
Prosodic modeling is a core problem in speech synthesis. The key challenge producing desirable prosody from textual input containing only phonetic information. In this preliminary study, we introduce the concept of "style tokens" Tacotron, recently proposed end-to-end neural synthesis model. Using style tokens, aim to extract independent prosodic styles training data. We show that without annotation data or an explicit supervision signal, our approach can automatically learn variety...
Adenosquamous cell carcinoma (ASC) is a rare and aggressive malignant tumor which consists of both adenocarcinoma (AC) squamous (SCC) component types. Although ASC can sometimes develop in the stomach, pancreas, gallbladder thyroid, it rarely occurs liver. As such, primary liver remains poorly understood malignancy due to paucity reported cases scarcity available published data. while use immune checkpoint inhibitors (ICIs), including PD-1 PD-L1 antagonists, has profoundly changed treatment...
The association of niacin intake with dyslipidemia remains uncertain. aim this study was to explore the between dietary and prevalence among adults in United States (US). Data were obtained from National Health Nutrition Examination Survey (NHANES) conducted 2005 2014. exposure variable intake, measured through 24-hour recall interviews treated as both a continuous categorical variable. Dyslipidemia, defined by diagnostic criteria, outcome. Logistic regression restricted cubic spline models...
Targeting Nicotinamide adenine dinucleotide (NAD) metabolism has emerged as a promising anti-cancer strategy; we aimed to explore the health benefits of boosting NAD levels with nicotinamide riboside (NR) on hepatocellular carcinoma (HCC). We established three in vivo tumor models, including subcutaneous transplantation model both Balb/c nude mice (xenograft), C57BL/6J (allograft), and hematogenous metastatic neoplasm mice. NR (400 mg/kg bw) was supplied daily gavage. In-situ growth or...
Neoadjuvant chemoimmunotherapy is an important therapeutic modality for resectable nonsmall cell lung cancer (NSCLC). The roles of the neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio, and lymphocyte-to-monocyte in predicting efficacy prognosis patients with NSCLC receiving neoadjuvant remain uncertain. This study aimed to explore association baseline preoperative NLR, treatment response survival treated chemoimmunotherapy.Data between May 2019 July 2022 at our institute,...
We propose an end-to-end trainable network that can simultaneously detect and recognize text of arbitrary shape, making substantial progress on the open problem reading scene irregular shape. formulate shape detection as instance segmentation problem; attention model is then used to decode textual content each irregularly shaped region without rectification. To extract useful features from image scale features, we a simple yet effective RoI masking step. Additionally, show predictions...
We explored the mechanism by which LINC00265 regulates angiogenesis of osteosarcoma cells via miR-382-5p/spermidine/spermine N1-acetyltransferase-1 (SAT1) and miR-382-5p/vav guanine nucleotide exchange factor 3 (VAV3) axis. Cell scratch assay, Transwell assay tube formation were applied to detect cell migration, invasion abilities. The effects targeting miR-382-5p in vivo studied using a tumour-burden assay. A total 70 genes potentially involved identified, Gene Ontology (GO) analysis found...
Programmed cell death-ligand 1 (PD-L1) expressed on cancer cells can cause immune escape of non-small-cell lung (NSCLC). Elucidation the regulatory mechanisms PD-L1 expression is a prerequisite for establishing new tumor immunotherapy strategies. Ubiquitin C-terminal hydrolase L1 (UCHL1) regulator cellular signaling transduction that aberrantly in NSCLC. However, it not known whether UCHL1 regulates NSCLC cells. In present study, we found promotes lines. addition, inhibited activation Jurkat...
Metastasis of hepatocellular carcinoma (HCC) can be facilitated by TNF-α, a prototypical inflammatory cytokine in the HCC microenvironment. A20 is negative regulator NF-κB signaling pathway. In present study we ask whether plays role metastasis. We found that expression was downregulated invasive cells microvascular invasions (MVI) compared with noninvasive 89 tissue samples from patients immunochemistry methods. Overexpression cell lines inhibited their motility induced TNF-α. Furthermore,...
Abstract Background and aims Hepatocellular carcinoma (HCC) with portal vein tumor thrombus (PVTT) has an extremely poor prognosis. A previous study proved that low-dose radiotherapy (RT) could prolong the prognosis of HCC patients PVTT. This to explore sensitivity PVTT RT treatment. Methods Patients were selected based on imaging diagnosis accompanied by received combined treatment radiotherapy, antiangiogenic drugs immune checkpoint inhibitors, followed hepatectomy or liver transplantation...
(Pers.) Roussel (1806) is a large genus of Mycenaceae known for having small to medium-sized basidiomata. It typified by the species
On the basis of studying datasets students' course scores, we constructed a Bayesian network and undertook probabilistic inference analysis. We selected six requisite courses in computer science as nodes. determined order nodes based on expert knowledge. Using 356 datasets, K2 algorithm learned structure. Then, used maximum posteriori probability estimation to learn parameters. After constructing network, message-passing predict infer results. Finally, results dynamic knowledge were...