Hello! I am Yue Zhou. Welcome to visit my Github Page.
Currently, I am working as an algorithm engineer at Peng Cheng Laboratory, Shenzhen, China.
My research interest including deep learning, distributed machine learning and high performance computing. Futhermore, deploying such algorithms into science, image processing and NLP domain.
April 2020 - Present
Design end-to-end AI benchmark suite that utilizing automated machine learning to evaluate large computer cluster.
Published Paper: AIPerf: Automated machine learning as an AI-HPC benchmark
Optimizing transformer-based model and applying it to gravitational wave data, achieving low SNR signal denoising and black binary hole parameter estimation.
Published Paper: Sampling with prior knowledge for high-dimensional gravitational wave data analysis
July 2019 - March 2020
In charge of the development and maintenance of blood analyzer, including:
Optimizing system and algorithm of the product;
Handling online issue of customer;
Pre-research of next-generation product.
June 2018 - August 2018
Implementation of multi-task convolutional neural network for face recognition and alignment and face quality assessment.
Implementation of face quality assessment algorithm.
June 2017 - June 2019
Training and fine-tuning discrete neural network to classify natural images and medical images (eye fundus images) based on Theano and Tensorflow framework, achieving high prediction accuracy and reducing memory/computation consumption.
Published Paper: Automatic Cataract Classification Using Deep Neural Network With Discrete State Transition
January 2016 - June 2017
Designed automatic ROI extraction algorithms of microfilament and nuclei images. Designed automatic nucleus detection and segmentation algorithms. Applying structure tensor method to compute orientation of microfilaments and nucleus automatically, and studying its relation with cell culture environment.
Published Paper: Automatic directional analysis of cell fluorescence images and morphological modeling of microfilaments
2016 - 2019
Beijing Institute of Technology
Major research domian includes Medical Image Processing, Machine Learning and Deep learning, follows Professor Huiqi Li.
I was awarded as excellent postgraduate and won National Scholarship.
My research thesis is ranked as outstanding achievement by Chinese Institute of Electronics.
2012 - 2016
Beijing Institute of Technology
Major courses include Digital Signal Processing, Communication Theory, Microwave Theory, Electronic Engineering and System,
Digital Image Processing, et. al.
I Graduated as distinguished graduate and recommended to continue post-graduate study.
Automatic Cataract Classification Using Deep Neural Network With Discrete State Transition
Automatic cataract detection and grading methods are proposed in this paper. With prior knowledge, the improved Haar features and visible structure features are combined as features, and multilayer perceptron with discrete state transition (DST-MLP) or exponential DST (EDST-MLP) are designed as classifiers. Without prior knowledge, residual neural networks with DST (DST-ResNet) or EDST (EDST-ResNet) are proposed.
Sampling with prior knowledge for high-dimensional gravitational wave data analysis
In this study, we incorporate prior physical knowledge by sampling from desired interim distributions to develop the training dataset. We adapt the normalizing flow method to be more expressive and trainable, such that the information can be effectively extracted and represented by the transformation between the prior and target distributions.
Cytoskeleton and nucleus are two important anatomic components in eukaryotic cells. Cell fluorescence images are employed to study their realignment and deformation during cell extrusion. Quantitative analysis and modeling of cell orientation are investigated in this paper.
AIPerf: Automated machine learning as an AI-HPC benchmark
Current popular AI benchmarks, such as MLPerf, have a fixed problem size and therefore limited scalability. To address these issues, we propose an end-to-end benchmark suite utilizing automated machine learning, which not only represents real AI scenarios, but also is auto-adaptively scalable to various scales of machines.
Automatic directional analysis of cell microscopy images
In this paper, methods to detect orientation of actin filament and nucleus automatically in microscopy images are proposed. In actin filament image, ROI region is obtained using morphological operator and gradient. Derivative-based method is employed to measure the direction of actin filaments in each subarea. In nucleus images, level set with intensity adjustment is applied to segment nuclei.
Institute
Peng Cheng Laboratory
Address
Dashi 1st Rd, Shenzhen, Guangdong, China
zhouy@pcl.ac.cn