Experiences

Senior Data Scientist, Inventec Inc., Taiwan, 2021/08/04 - 

• AI Research:

  • Trustworthy AI: Develop secure AI inference with crypto technologies, e.g., FHE, MPC, or Hashing, and improve AI reliability

  • Knowledge Preservation: Develop a training framework to store learning knowledge with a low-dimensional data representation during ML training

• Research Management: Lead the direction of the department's AI research and maintain shared resources

• Improve AI literacy: Run a series of talks/lecture/hackathon to improve AI literacy in the company

• Execute CV-related Project: Help with designing solutions for CV-related projects, including Anomaly detection, defect detection, and long-term surveillance

Postdoc, Security and Systems Lab, National Yang Ming Chiao Tung University, 2021/04 - 2021/08

• Secured NN Inference for Malware Detection with data hashing

• Privacy-preserving Federated Learning with Multimodal Inputs

• Point cloud compression benchmarking

Research Assistant, Networking and Multimedia Systems Lab, National Tsing Hua University, 2021/03 - 2021/04

• Introducing CNN and RL models to address the configuration optimization problem in the 6DoF scenario

Postdoc, Computer Network and Machine Learning Lab, University of California, Davis, 2020/03/15 - 2021/03/14

• Evaluate Interactive Real-time System: Develop an evaluation method and metrics to investigate the Pros & Cons for Real-Time Communication System

• Adaptive Software Configuration: Investigate the RL algorithm to control the video encoder to find the optimal encoding configuration

Research Assistant, Networking and Multimedia Systems Lab, National Tsing Hua University, 2020/01/1 - 2020/02/29

• Propose a reinforcement model to select critical views for view synthesis to reduce the transmission rate and the synthesizing time

• PCC Benchmark: Develop a benchmark for evaluating heterogeneous Point Cloud Compression methods

Research Assistant, DIRL (MMNET) Lab, Academia Sinica, 2014/02 - 2019/12

• Propose a real-time gaze correction system based on a convolutional neural network to solve the missing eye contact problem for improving the engagement of the live video communication

• Propose a method to evaluate the performance of the screencast technologies and identify a few critical open problems for future design

• Design a framework to evaluate the efficiency and consistency of quality assessment methods

• Analyze the relationship between human perception and the settings of foveated images

Research Assistant, IPCV-ITS Lab, National Taiwan Normal University, 2011/09 - 2013/12

• Reconstruct 3D information of the aerial image to eliminate the blind spots surrounding a car to improve safety

• Improve the efficiency of salient object detection based on applying the branch-and-bound algorithm to speed up the sliding window search

2025/12/10, MMAsia 2025, x-Models in Manufacturing: Impressive, But Are They Ready?

2025/10/01 NKUST, 從電玩迷到企業資料科學家

2025/03/17, GTC2025, Accelerate FHE-Based DNN Secure Inference With GPU [P74396] (poster)

2024/03/06, National Formosa University Seminar, A brief Introduction of Generative AI

2022, 2023, 2024 National Taiwan University, Trustworthy AI (Elective courses co-instructor)

2023/03/23, GTC2023, Utilizing Data with Low Privacy Concerns in Smart Manufacturing (slides)

2022/04/22, National Taiwan University Seminar, Improving existing Systems with Machine Learning Approaches: Challenges and Opportunities

2021/11/26, National Formosa University Seminar, Video Transmission Systems Meet Machine Learning

2021/07/13-2021/07/27, National Taiwan Normal Univresity, 資訊科技第二專長學分班人工智慧課程

2021/06/02, National Taiwan Normal University Seminar, Video Transmission Systems Meet Machine Learning [pdf]

2020/11/26, 清華大學資工所 IoT-webinar, Real-time Communication: Opportunities and Challenges

2019/10/22, 2019/10/29, 2019/11/05, 成大巨資中心, 機器學習實務 (1)(2)(3)

2019/09/03, 清華大學資工所, Toward User Immersed Multimedia Applications: Studies on Gaze Correction in Live Video Communication

2018/11/12, 台灣大學資工所 Seminar, Look at Me! Correcting Eye Gaze in Live Video Communication

2018/07/17, 清華大學資工所, Flx-gaze: A Real-time Gaze Redirection System based on Warping-based CNN Model

2018/03/24, 逢甲大學資工所, 走馬看花的深度學習地城之旅 

2017/11/09, 台灣人工智慧年會, 當人臉遇到機器學習 [pdf]