Tzu-Hung Liang

• Develop sensor algorithms and evaluation workflows for wearable biosignal and industrial sensing applications. • Designed and implemented signal processing and evaluation methods for long-duration biosignal analysis, including multi-signal synchronization, intra-device channel integration, and interpretable reporting workflows. • Built reproducible evaluation environments by organizing algorithm repositories and providing Docker-based execution workflows and documentation for cross-functional teams between BUs. • Supported cross-generation hardware evaluation by modifying generation-specific thresholding and normalization strategies, enabling more consistent comparison across device versions. • Developed analysis and visualization tools for long-duration signal review, improving result interpretability and supporting hardware evaluation and planning discussions. • Evaluated anomaly detection robustness for industrial photoelectric sensing under varied physical environments and explored robust feature combinations through Python-based analysis.

- Researched advanced AI algorithm solutions for blood pressure estimation. - Developed a novel deep learning architecture to improve performance by over 20% and outperform the previous best method developed by the team. - Increased 25% of the efficiency for blood pressure estimation by the proposed algorithm. - Mitigated the overfitting problem with less than 300 samples in the dataset by feature engineering. - Conduct technical discussions in a full English environment.

- Engaged in a temporary research position at the lab of Professor Mehdi Norouzi at the University of Cincinnati. - Preprocess images/video frames and research their characteristics to extract an optimal input for deep learning models. - Researched and implemented various video summarization techniques, both traditional and state-of-the-art. - Reproduced state-of-the-art deep learning video summarization models for further investigation and improvement.

- Assembled and tested PCB machines. - Assisted in wiring and layout planning for PCB machines. - Study the new way to build connections between PLCs.

Courses taken: 1) Advanced Digital Image Processing 2) Pattern Recognition 3) Stochastic Processes 4) Artificial Intelligence and Machine Learning 5) Multi-Dimensional Image Processing 6) Building Deep Learning Applications 7) Advanced Object-Oriented Programming in Python