A5018240777- Speech Recognition and Synthesis
- Speech and Audio Processing
- Music and Audio Processing
- Crystallization and Solubility Studies
- X-ray Diffraction in Crystallography
- Solar Radiation and Photovoltaics
- Energetic Materials and Combustion
- Crystallography and molecular interactions
- Speech and dialogue systems
- Fluid Dynamics and Turbulent Flows
- Meteorological Phenomena and Simulations
- Energy Load and Power Forecasting
- Chemical Thermodynamics and Molecular Structure
- Thermal and Kinetic Analysis
- Metal Extraction and Bioleaching
- Photovoltaic System Optimization Techniques
- Atmospheric aerosols and clouds
- Climate variability and models
- Solar Thermal and Photovoltaic Systems
- Wind and Air Flow Studies
- Plant Water Relations and Carbon Dynamics
- Natural Language Processing Techniques
- Impact of Light on Environment and Health
- Advanced Memory and Neural Computing
- Quantum-Dot Cellular Automata
University of Electronic Science and Technology of China
2024
Purdue University West Lafayette
2024
University of Shanghai for Science and Technology
2024
China Academy of Engineering Physics
2023
University of South China
2016-2023
Jilin University
2023
Peking University
2023
University of Oxford
2023
Sichuan University
2020-2022
Beijing Jiaotong University
2017-2022
Deep belief networks (DBN) have shown impressive improvements over Gaussian mixture models for automatic speech recognition. In this work we use DBNs audio-visual recognition; in particular, deep learning from audio and visual features noise robust We test two methods using a multimodal setting: conventional decision fusion method that combines scores single-modality DBNs, novel feature operates on mid-level learned by the DBNs. On continuously spoken digit recognition task, our experiments...
Transition-metal-free trifluoromethylations have attracted significant research interest driven by the increasing importance of CF 3 -containing compounds.
A simple and practical electron donor-acceptor (EDA) strategy to synthesize various 3-alkylated coumarins from easily available naturally abundant carboxylic acids under photocatalyst-, oxidant-, additive-free mild conditions is reported. Using Na2S as the catalytic donor, a series of primary, secondary, tertiary carbon radicals can be efficiently generated, EDA complex regenerated without an alkaline additive.
The proliferation of street view images (SVIs) and the constant advancements in deep learning techniques have enabled urban analysts to extract evaluate perceptions from large-scale streetscapes. However, many existing analytical frameworks been found lack interpretability due their end-to-end structure "black-box" nature, thereby limiting value as a planning support tool. In this context, we propose five-step machine framework for extracting neighborhood-level panoramic SVIs, specifically...
There has been considerable research on quantum dot cellular automata (QCA) as a new computing scheme in the nano-scale regimes. The basic logic element of this technology is majority voter. In paper, detailed simulation-based characterization QCA defects and study their effects at logic-level are presented. Testing these devices investigated compared with conventional CMOS-based designs. Unique testing features designs based presented interesting properties have identified.
Abstract This study seeks to quantitatively and qualitatively understand how stability affects transport in the continuously turbulent stably stratified atmospheric boundary layer, based on a suite of large-eddy simulations. The test cases are one adopted by Global Energy Water Cycle Experiment (GEWEX) Atmospheric Boundary Layer Study (GABLS) project, but with largely expanded range where gradient Richardson number (Rig) reaches up around 1. analysis is mainly focused understanding...
Defect tolerance is an extremely important aspect in nano-scale electronics as the bottom-up self-assembly fabrication process results a significantly higher defect density compared to conventional lithography-based processes. techniques are therefore essential obtain acceptable manufacturing yield. In this paper, we investigate properties of 2D crossbar, which basic block various nano architectures have been recently proposed. Various nano-wire and switch faults studied their impact on...
Current hidden Markov acoustic modeling for large-vocabulary continuous speech recognition (LVCSR) heavily relies on the availability of abundant labeled transcriptions. Given that labeling is both expensive and time-consuming while there a huge amount unlabeled data easily available nowadays, semi-supervised learning (SSL) from aiming to reduce development cost LVCSR becomes more important than ever. In this paper, new SSL approach proposed which exploits cross-view transfer through...
This paper discusses the differences, due to different measuring mechanisms, of satellite-derived and ground measured surface solar irradiance (SSI) direct diffuse components, impacts differences when validating data using measurements, SSI products derived from Advanced Himawari Imager. The issue satellite-ground mismatch, though previously recognized, has not been given sufficient attention in relevant scientific literature to: (i) satellite have mostly focused on relatively longer-term...
This paper is about an algorithm, FlexTree, for general supervised learning. It extends the binary tree-structured approach (Classification and Regression Trees, CART) although it differs greatly in its selection combination of predictors. particularly applicable to assessing interactions: gene by environment as they bear on complex disease. One model predisposition disease involves many genes. Of them, most are pure noise; each values that not prevalent genotype minority genes contribute...