About

I am currently a Phd candidate of the Hong Kong Polytechnic University (PolyU), supervised by Prof. Kit Lun Yick in AiDLAB. Before that, I was a machine learning engineer at XiaoMi Inc. I received my Master degree from the Department of Computer Science, Beijing Jiaotong University, under the supervision of Prof. Jing Wang. My research interests include 3D Vison, Anomaly Detection, Self-supervised / Unsupervised Learning and Deep Learning Methods for Physiological Time-series. Recently I am focusing on learning semantic representations for physiological time-series.

Work Experience

1. Machine Learning Engineer - XiaoMi Inc. (2021 - 2023)

I’m responsible for developing state-of-the-art algorithms of real-world problems applied on wearable devices, including:

  • energy expenditure prediction based on Transformer and multi-task learning;

  • Xiaomi IoT application: smart TV controlling using wearable devices and gesture recognition;

  • smart watch fall detection with wearable device sensors, e.g. accelerator and gyroscope;

  • smart fitness assessment based on collected physilogical data, e.g. heartrate, breathing rate, pressure and tracking
    data;

Projects

PyADTS (2020)

PyADTS is aimed at accelerating the workflow of time series anomaly detection for researchers. It contains various modules for data loading, pre-processing, anomaly detection, detector ensembling, evaluation and etc.

Link: https://github.com/larryshaw0079/PyADTS

Publications

  • Jianan Ye*, Qinfeng Xiao*. CoSleep: A Multi-view Representation Learning Framework for Self-Supervised Learning of Sleep Stage Classification. IEEE Signal Processing Letters, 2021. (* equal contribution) [Code | Paper]

  • Qinfeng Xiao. Self-Supervised Learning for Sleep Stage Classification with Predictive and Discriminative Contrastive Coding. The International Conference on Acoustics, Speech, & Signal Processing (ICASSP), 2021. [Code | Paper]

  • Qinfeng Xiao. Unsupervised Anomaly Detection with Distillated Teacher-Student Network Ensemble. Entropy, 2021. [Code | Paper]

  • Yunxiao Liu, Youfang Lin, Qinfeng Xiao, Ganghui Hu, Jing Wang. Self-adversarial Variational Autoencoder with Spectral Residual for Time Series Anomaly Detection. Neurocomputing, 2021. [Paper]

  • Qinfeng Xiao, Shikuan Shao, Jing Wang. Memory-augmented Adversarial Autoencoders for Multivariate Time-series Anomaly Detection with Deep Reconstruction and Prediction. ArXiv, 2021. [Paper]

  • Jing Tang, Qinfeng Xiao, Zhiguo Gui, Baosheng Li, Pengcheng Zhang. Simulation of Proton-Induced DNA Damage Patterns Using an Improved Clustering Algorithm. Radiation Research, 2020. [Paper]