My name is Wei Pengfei. I am currently a research scientist in the AI lab of Bytedance. Our team is recruiting new members, and welcome to join us! All the positions are available here.

Before I joined Bytedance, I worked as a research fellow in the Medcomp Lab of National University of Singapore from May 2019 to March 2021 (advisor: Prof. Leong Tze Yun). I recieved my PhD degree from the School of Computer Science and Engineering, Nanyang Technological University, Singapore, in 2019. My PhD supervisor is Prof. Ke Yiping.

Research Interests

My research interests inculde transfer learning, domain adaptation, reinforcement learning and their interdisciplinary topics as well as thier applications to computer vision, natural langauge process, speech signal processing and system engineering etc.

Recent News

• The paper titled “Prior and Prediction Inverse Kernel Transformer for Single Image Defocus Deblurring” is accepted by AAAI 2024.

• The paper titled “Unsupervised Video Domain Adaptation for Action Recognition: A Disentanglement Perspective” is accepted by NeurIPs 2023.

• The paper titled “Virtual Try-On with Pose-Garment Keypoints Guided Inpainting” is accepted by ICCV 2023.

• The paper titled “S2CD-VC: Self-heuristic Speaker Content Disentanglement for Any-to-Any Voice Conversion” is accepted by Interspeech 2023.

• The paper titled “AudioQR: Deep Neural Audio Watermarks For QR Code” is accepted by IJCAI 2023.

• The paper titled “Adaptive Policy Learning for Offline-to-Online Reinforcement Learning” is accepted by AAAI 2023.

• The paper titled “Adaptive Transfer Kernel Learning for Transfer Gaussian Process Regression” is accepted by TPAMI.

• The paper titled “Transfer Kernel Learning for Multi-Source Transfer Gaussian Process Regression” is accepted by TPAMI.

Selected Papers

• Pengfei Wei, Yiping Ke, Xinghua Qu, Tze Yun Leong. “Subdomain Adaptation with Manifolds Discrepancy Alignment”, IEEE Transactions on Cybernectics, 2021.

• Pengfei Wei, Ramon Sargana, Yiping Ke, Yew Soon Ong. “Practical Multisource Transfer Regression with Source–Target Similarity Captures”, IEEE Transactions on Neural Networks and Learning systems, 2020.

• Pengfei Wei, Yiping Ke. “Knowledge Transfer based on Multiple Manifolds Assumption”. International Conference on Information and Knowledge Management (CIKM), 2019, Beijing, China, 279-287.

• Pengfei Wei, Yiping Ke, Chi Keong Goh. “A General Domain Specific Feature Transfer Framework for Hybrid Domain Adaptation”, IEEE Transactions on Knowledge and Data Engineering 31.8 (2018): 1440-1451.

• Pengfei Wei, Yiping Ke, Chi Keong Goh. “Domain Specific Feature Transfer for Hybrid Domain Adaptation”, International Conference on Data Mining (ICDM), 2017, 1027-1032.

• Pengfei Wei, Yiping Ke, Chi Keong Goh. “Deep Nonlinear Feature Coding for Unsupervised Domain Adaptation”, International Joint Conference on Artificial Intelligence (IJCAI), 2016, 2189-2195.