Yueh-Po Peng (Yabi)

Yueh-Po Peng

About Me

I am a Research Assistant at the Institute of Information Science, Academia Sinica, focusing on self-supervised learning methods for decoding mental states from brain activity using fMRI data. I have a background in Data Science and Computer Science, with extensive experience in machine learning, neuroscience, and music information research.

Experience

• Institute of Information Science, Academia Sinica, Taipei, Taiwan
  Research Assistant (Jul. 2024 – Present)
  • Surveyed end-to-end self-supervised learning methods for decoding mental states from brain activity (fMRI).
  • Conducted distributed training experiments on large-scale, high-resolution 4D fMRI data using TWCC HPC.
• Tomofun, Taipei, Taiwan
  Research & Development - AI Intern (Mar. 2023 – Jul. 2024)
  • Developed an automatic short music video generation system for daily pet clips.
  • Surveyed various strategies of visual language models (LLaVA) to generate image-caption pairs for knowledge distillation.
• Institute of Information Science, Academia Sinica, Taipei, Taiwan
  Research Assistant (Mar. 2022 – Feb. 2023)
  • Proposed a whole-brain feature selection method for decoding musical pitch from brain activity (fMRI).

Education

• National Taiwan University
  M.S. in Data Science (Feb. 2023 – Jun. 2024)

• National Taiwan University
  B.S. in Computer Science and Information Engineering (Sep. 2019 – Jan. 2022)

Research & Projects

Guitar Effect Removal

Machine Learning Research on Removing Distortion Effect from Electric Guitar

• Proposed a two-stage method to remove distortion effects from guitar recordings using Positive Grid VST plugins.
• Analyzed baseline models on synthetic and VST-rendered effects, demonstrating superior performance.
• Published in DAFx 2024. Paper, Demo

Whole Brain fMRI Features Selection

Machine Learning Research to Find Correlation Between fMRI and Musical Pitch

• Proposed a two-stage method to extract fMRI features and predict musical pitch.
• Evaluated ML models’ performance and analyzed correlation between pitch and fMRI patterns.
• Published in ICASSP 2023. Paper

Publications

1. Ying-Shuo Lee*, Yueh-Po Peng*, Jui-Te Wu, Ming Cheng, Li Su, and Yi-Hsuan Yang (2024).
   “Distortion recovery: A two-stage method for guitar effect removal.”
   In Proc. Int. Conf. Digital Audio Effects (DAFx), 2024. (* equally contributed)

2. Cheung, V. K.*, Peng, Y. P.*, Lin, J. H., & Su, L. (2023, June).
   “Decoding Musical Pitch from Human Brain Activity with Automatic Voxel-Wise Whole-Brain FMRI Feature Selection.”
   In ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), (pp. 1-5). IEEE. (* equally contributed)

Skills

• Languages/Frameworks: Python, PyTorch, TensorFlow, Pandas, Scikit-learn, Slurm, Flask, HTML, JavaScript, C++, C, Linux
• Skillset: Machine Learning, Self-Supervised Learning, Neuroscience, Music Information Research, Distributed Training