Work Experience2016 ~ Now• Proposed the Emotion Modulated Harmonics(EMH) model and firstly implemented the ANN-HMM framework for continuous emotion recognition with traditional physiological signals and optical facial blood flow signals.• Proposed the Stationary Harmonic Response (SHR) model and got the world-leading performance of cuff-free photoplethysmogram (PPG) based blood pressure measurement by combining the SHR model with the deep learning DNN/RNN-LSTM model.• Proposed the Blood Volume Grey Modulation (BVGM) model and the 3DWF-3DCNN-ResNet deep learning model, based on which developed a cuff-free blood pressure measurement system with the common camera collected optical facial blood flow signals.2003 ~ 2015• Got the world top level performance in the large vocabulary continuous Mandarin speech recognition with Toshiba by using the unique pitch extraction algorithm, the tonal acoustic modeling with discriminative training technology, the novel new words and key words registration algorithm, the ANN-HMM hybrid acoustic model and the ARPA-RNN hybrid language model.• Proposed a bunch of noise robust speech recognition technologies and helped Toshiba win the DENSO competitions of embedded in-car speech recognition.Highlights of Profiles• Have over 16 years experience in the field of speech recognition and machine learning.• Experienced in deep learning and machine learning algorithms, especially neural network technologies.• Have broad research interests in speech enhancement, speech signal processing, nature language processing, pattern recognition and artificial intelligence.• Have the outstanding knowledge and the solid theoretical foundation in mathematics， physics, probability theory and information theory.• Have the project management experience in the R&D of speech recognition technologies and systems.• Excellent communication and leadership skills with ability to build an efficient team.• Familiar with C, Perl, Python, Matlab and Linux.