GUOXING
GUOXING
≡
profile photo

Guoxing Lan

I am currently a senior algorithm engineer at miHoYo, where I focus on NLP (Natural Language Processing). Before that, I was a senior algorithm engineer at Huawei Consumer Business Group, Shanghai, where I worked on applications and pre-research of machine learning/deep learning (mainly for Huawei's voice assistant Xiaoyi).
I obtained a master's degree and a bachelor's degree both from Department of Automation, and a second bachelor's degree in economics from School of Economics and Management, in Tsinghua University.
I enjoy writing blogs( CSDN Blog ) and push codes to my github, because I believe the best way to learn new knowledge is by explaining it clearly to other people and utilizing it to solve practical problems. Besides, I think sharing knowledge with others is one of the most wonderful things in the world!

Email  /  LinkedIn  /  Google Scholar  /  Github  /  CSDN Blog  /  CV

News

  • 02/2022, I left Huawei CBG and joined miHoYo as a senior algorithm engineer.

  • 05/2020, I changed my workplace from Beijing to Shanghai.

  • 08/2018, I joined Huawei CBG and worked as an AI algorithm engineer in Beijing.

  • 07/2018, I graduated from Tsinghua University with a master's degree.

  • 07/2015, I graduated from Tsinghua University with a bachelor's degree in Automation and a second bachelor's degree in Economics.

    • Selected Projects

    I have done several projects based on machine learning/deep learning methods.

  • Personalized ASR Algorithms and Application

    • I lead a team of 4 members doing pre-research on personalized ASR algorithms and application. We design and implement an architecture to fine-tune a pre-trained ASR model for each user with accent. Our aim is to improve the performance of the target user without worsening the performance of normal users. My major responsibility is to design the whole architecture and experiment plans, implement the main program and some crucial modules such as model evaluation module. During this project, I proposed a novel method to alleviate the knowledge forgetting problem when fine-tuning classification models (with softmax layer) with new personalized samples. One relevant patent has been applied by me.

  • Miniaturization and Acceleration of AI models

    • I focused on the miniaturization and acceleration of AI models to reduce the consumption of ROM, RAM and power, and increase the inference speed, with little or no accuracy loss. Four relevant patents has been applied by me.

  • NLU of Huawei’s Voice Assistant

    • I was responsible for the domain classification module and intent re-ranking module of the NLU (Natural Language Understanding) system of Huawei’s voice assistant (Xiaoyi). Xiaoyi has served hundreds of millions of smart phone users. One revelant patent was applied y me and has been published.

  • Remaining Useful Life Estimation of Turbofan Engine Using LSTM Neural Networks

    • Project Page | Paper

    The aim is to estimate the remaining useful life (RUL) of turbofan engine with degradation. I proposed a novel data-driven method based on LSTM neural networks to estimate the RUL with multivariate outputs of sensors and operational settings.

    Talks

  • 11/2018, "Introduction to Gradient Boosting Decision Trees", inside Huawei Company.

  • 11/2020, "Introduction to Text Classification", inside Huawei Company.

  • 05/2021, "Introduction to GPU Operators in Tensorflow-Lite", inside Huawei Company.

  • 10/2021, "Significance Analysis of AI Models Before and After Launching", inside Huawei Company.

  • 02/2022, "Introduction to CSI (Channel State Information) and Copresence Detection", inside Huawei Company.

  • 04/2022, "Model Compression and Acceleration Techniques in Deep Learning", inside miHoYo Company.

    • Research

    My interests are in Machine Learning and Deep Learning theories and their applications especially in NLP(Natural Language Processing). Besides, I have some research experiences in data-driven fault diagnosis and remaining useful life estimation during my master study.

    Research Experience:

  • Department of Automation , Tsinghua University Sep. 2016 – Feb. 2018
    • Research Assistant; Supervisor: Prof. Nong Cheng and Prof. Qing Li
    I did research on modeling and simulation of turbofan engine, data-driven fault diagnosis and remaining useful life estimation.

  • Department of Electrical and Computer Engineering, University of Alberta 2017 – Nov. 2017
    • Research Intern; supervisor: Prof. Venkata Dinavahi
    I designed and implemented PSO algorithms for nonlinear constrained optimization problems.

    Conference Papers:

    1. Remaining Useful Life Estimation of Turbofan Engine Using LSTM Neural Networks
      Guoxing Lan, Qing Li, Nong Cheng
      CGNCC 2018 | paper

    2. Comparison and fusion of various classification methods applied to aero-engine fault diagnosis
      Guoxing Lan, Nong Cheng, Qing Li
      CCDC 2017 | paper

    Services

  • Session Chair of CCDC 2017.

    • Awards

  • 2015, Excellent graduation thesis in Economics (For Second Bachelor Degree), School of Economics and Management, Tsinghua University

  • 2013, HAGE Scholarship, Department of Automation, Tsinghua University

  • 2012, HAGE Scholarship, Department of Automation, Tsinghua University

    • Leisure Interests

  • Writing technology blogs, pushing codes to GitHub, and sharing knowledge

  • Physical exercise including strength and aerobic training

    • The template of this homepage is from Yunhe Wang. ✩
    Aloha! You're the No. Visitor Since Oct 17, 2021. Powered by w3.css View My Stats