A5100628448- Speech Recognition and Synthesis
- Speech and Audio Processing
- Music and Audio Processing
- Geotechnical Engineering and Analysis
- Geotechnical Engineering and Soil Stabilization
- Geotechnical Engineering and Underground Structures
- Metallurgy and Material Forming
- Advanced Photocatalysis Techniques
- International Business and FDI
- Laser and Thermal Forming Techniques
- Concrete and Cement Materials Research
- Geotechnical Engineering and Soil Mechanics
- Fusion materials and technologies
- Innovative concrete reinforcement materials
- RNA modifications and cancer
- Structural Load-Bearing Analysis
- Domain Adaptation and Few-Shot Learning
- Advanced Image and Video Retrieval Techniques
- Metal Forming Simulation Techniques
- Face recognition and analysis
- Advanced Algorithms and Applications
- Mechanical stress and fatigue analysis
- Layered Double Hydroxides Synthesis and Applications
- Metal and Thin Film Mechanics
- EEG and Brain-Computer Interfaces
Jinan University
2024
Shenyang Jianzhu University
2020-2024
Xuzhou Medical College
2023-2024
North China University of Technology
2024
Xinjiang Academy of Agricultural Sciences
2023
Jilin University
2016-2022
Union Hospital
2018-2022
Central South University
2021-2022
BD Biosciences (United States)
2022
Nanjing University of Aeronautics and Astronautics
2022
Scene Graph Generation (SGG) has attracted more and attention from visual researchers in recent years, since (SG) is valuable many downstream tasks due to its rich structural-semantic details. However, the ap-plication value of SG on severely lim-ited by predicate classification bias, which caused long-tailed data presented as semantic bias predicted relation predicates. Existing methods mainly reduce prediction better aggregating contexts integrating external priori knowledge, but rarely...
Multimodal knowledge graph completion (MKGC) aims to predict missing entities in MKGs. Previous works usually share relation representation across modalities. This results mutual interference between modalities during training, since for a pair of entities, the from one modality probably contradicts that another modality. Furthermore, making unified prediction based on shared treats input different equally, while their importance MKGC task should be different. In this paper, we propose MoSE,...
Background Increasing evidence has suggested that major depression ( MD ) is associated with an increased risk of ischemic heart disease IHD ). We examined this association in Chinese adults using data from the China Kadoorie Biobank study. Methods and Results Over 0.5 million aged 30 to 79 years were followed baseline interview (2004–2008) until December 31, 2013. Past year was measured modified version Composite International Diagnostic Interview‐Short Form at baseline. Incident cases...
Mechanical reinforcement by plant roots increases the soil shearing strength. The geometric and distribution characteristics of affect Current research on shear strength rooted‐soil is mostly based direct tests with a fixed surface thus cannot reflect actual failure state rooted‐soil. In this study, Golden Vicary Privet was used to create rooted‐soil, triaxial test method for mechanical property analysis. influence root geometry (root diameter individual length) density angle) studied...
Ink based on nanoconductive materials can achieve the scalable fabrication of flexible devices. Nanocarbon materials, such as reduced graphene oxide (rGO), (GO), and carbon nanotubes (CNT), offer great potential in field electronics. However, these are challenging to use inks, especially environmentally friendly water–soluble due their poor dispersion most solvents. In this study, a biocompatible well-dispersed keratin-based rGO ink was fabricated consisting keratin extracted from waste wool...
In this paper, we introduce a large-scale and high-quality audiovisual speaker verification dataset, named VoxBlink. We propose an innovative robust automatic audio-visual data mining pipeline to curate which contains 1.45M utterances from 38K speakers. Due the inherent nature of automated collection, introducing noisy is inevitable. Therefore, also utilize multi-modal purification step generate cleaner version VoxBlink, VoxBlink-clean, comprising 18K identities 1.02M utterances. contrast...
Identification of individuals is ubiquitous with increasing reliance by financial and governmental organizations on reliable robust personal recognition systems to determine confirm the identity policy constraints for specific in 'real-time' when reacting service requests. The traditional identification approaches user validation are not deficiencies such becoming increasingly apparent current information-oriented society. With developments research into human brain, biometric methods based...
Purpose: To explore the role of Th2 signaling pathway in allergic conjunctivitis (AC). Methods: Serum cytokines IL-4 or IL-13 patients with AC were detected using Meso scale discovery assay to verify correlation immunity and pathogenesis. Wistar Han rats intraperitoneally subcutaneously injected ovalbumin (OVA) establish an experimental model was blocked by investigational neutralizing antibody (CM310). IgE OVA-specific ELISA. Conjunctivitis inflammation, infiltration eosinophils, mast cell...
Abstract Cancer metabolism has emerged as a major target for cancer therapy, while the state of mitochondrial drugs remained largely unexplored, partly due to an inadequate understanding various functions in tumor contexts. Here, we report that HOMER3 is highly expressed non-small cell lung (NSCLC) and closely correlated with poor prognosis. Lung cells low levels are found show significant dysfunction, thereby suppressing their proliferation metastasis vivo vitro. At mechanistic level,...
Recent approaches have demonstrated the effectiveness of Vision Transformer (ViT) with attention mechanisms for domain generalization Face Anti-Spoofing (FAS). However, current algorithms highlight all salient objects (e.g., background objects, hair, glasses), which results in feature learned by model containing face-irrelevant noisy information. Inspired existing Vision-language works, we propose VL-FAS to extract more generalized and cleaner discriminative features. Specifically, leverage...
Lung cancer is the most common malignant tumor. In present study, we identified a long non-coding RNA (lncRNA) AC100826.1 (simplify to Lnc1), which was highly expressed in non-small cell lung (NSCLC) tissues compared with paracancerous tissues. We also observed critical role of Lnc1 regulating metastasis ability NSCLC cells.
Soil deformation control is the key to shaft support. To better soil deformation, improve construction efficiency, and reduce pollution, this study proposed a prefabricated prestressed supporting structure. The structure consisted of steel units special components. were applied retain soil. screws used for prestressing. Field prototype tests conducted assess support effects analyze stress behaviors shaft. earth pressure, in unit, lateral displacement monitored. measured pressure varied...
End-to-end automatic speech recognition (ASR) usually suffers from performance degradation when applied to a new domain due shift. Unsupervised adaptation (UDA) aims improve the on unlabeled target by transferring knowledge source domain. To transferability, existing UDA approaches mainly focus matching distributions of and domains globally and/or locally, while ignoring model discriminability. In this paper, we propose novel approach for ASR via inter-domain MAtching intra-domain...