Publications
Multimodal Large Language Models (MLLMs)
COMPBENCH: A Comparative Reasoning Benchmark for Multimodal LLMs.[ pdf ]
Zheda Mai*, Jihyung Kil*, Justin Lee, Zihe Wang, Kerrie Cheng, Lemeng Wang, Ye Liu, Arpita Chowdhury, Wei-Lun Chao
In Proceedings of the Conference on Neural Information Processing Systems (NeurIPS), 2024
Efficient Foundation Model Adaptation
Lessons Learned from a Unifying Empirical Study of Parameter-Efficient Transfer Learning (PETL) in Visual Recognition.[ pdf ]
Zheda Mai, Ping Zhang, Cheng-Hao Tu, Hong-You Chen, Li Zhang, Wei-Lun Chao
Under Review, 2024
Fine-Tuning is Fine, if Calibrated.[ pdf ]
Zheda Mai* ,Arpita Chowdhury*, Ping Zhang*, Cheng-Hao Tu, Hong-You Chen, Vardaan Pahuja, Tanya Berger-Wolf, Song Gao, Charles Steward, Yu Su, Wei-Lun Chao
In Proceedings of the Conference on Neural Information Processing Systems (NeurIPS), 2024
Holistic Transfer: Towards Non-Disruptive Fine-Tuning with Partial Target Data. [ pdf ]
Cheng-Hao Tu*, Hong-You Chen*, Zheda Mai, Jike Zhong, Vardaan Pahuja, Tanya Berger-Wolf, Song Gao, Charles Steward, Yu Su, Wei-Lun Chao
In Proceedings of the Conference on Neural Information Processing Systems (NeurIPS), 2023
Visual Query Tuning: Towards Effective Usage of Intermediate Representations for Parameter and Memory Efficient Transfer Learning [ pdf ]
Zheda Mai*, Cheng-Hao Tu*, Wei-Lun Chao
In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023
Continual Learning
Online Class-Incremental Continual Learning with Adversarial Shapley Value [ pdf ] [ Presentation ] [ code ]
Zheda Mai*, Dongsub Shim*, Jihwan Jeong*, Scott Sanner, Hyunwoo Kim and Jongseong Jang
In 35th AAAI Conference on Artificial Intelligence (AAAI), 2021
Supervised Contrastive Replay: Revisiting the Nearest Class Mean Classifier in Online Class-Incremental Continual Learning [ pdf ] [ code ]
Zheda Mai, Ruiwen Li, Hyunwoo Kim, Scott Sanner
In Workshop on Continual Learning in Computer Vision at Conference on Computer Vision and Pattern Recognition (CVPR), 2021
CVPR 2020 Continual Learning in Computer Vision Competition: Approaches, Results, Current Challenges and Future Directions [ pdf ]
Vincenzo Lomonaco, ... Zheda Mai, etc.
Artificial Intelligence Journal (AIJ), 2021
Batch-level Experience Replay with Review for Continual Learning [ pdf ] [ code ] [ Presentation ]
Zheda Mai, Hyunwoo Kim, Jihwan Jeong, Scott Sanner
In Workshop on Continual Learning in Computer Vision at Conference on Computer Vision and Pattern Recognition (CVPR), 2020
Weakly Supervised Learning
Segment Anything Model (SAM) Enhances Pseudo Labels for Weakly Supervised Semantic Segmentation [ pdf ]
Zheda Mai*, Tianle Chen*, Ruiwen Li, Wei-Lun Chao
Preprint, 2023.
TransCAM: Transformer Attention-based CAM Refinement for Weakly Supervised Semantic Segmentation [ pdf ]
Ruiwen Li, Zheda Mai, Chiheb Trabelsi, Zhibo Zhang, Jongseong Jang, Scott Sanner
In Journal of Visual Communication and Image Representation, 2023.
Recommender Systems
Towards understanding and mitigating unintended biases in language model-driven conversational recommendation
Tianshu Shen, Jiaru Li, Mohamed Reda Bouadjenek, Zheda Mai, Scott Sanner
In Information Processing & Management (IPM), 2023.
Mitigating the Filter Bubble while Maintaining Relevance: Targeted Diversification with VAE-based Recommender Systems
Zhaolin Gao, Tianshu Shen, Zheda Mai, Mohamed Reda Bouadjenek, Isaac Waller†, Ashton Anderson, Ron Bodkin, Scott Sanner
In Proceedings of the 45th international ACM SIGIR conference on Research and development in information retrieval (SIGIR), 2022.
Distributional Contrastive Embedding for Clarification-based Conversational Critiquing [pdf]
Zheda Mai*, Tianshu Shen*, Ga Wu, Scott Sanner
In Proceedings of the ACM Web Conference (WWW), 2022.
Attentive Autoencoders for Multifaceted Preference Learning in One-class Collaborative Filtering [pdf] [ Presentation ]
Zheda Mai*, Ga Wu*, Kai Luo, Scott Sanner
In Workshop on Advanced Neural Algorithms and Theories for Recommender Systems (NeuRec) at IEEE International Conference on Data Mining (ICDM), 2020. (Oral).
Noise Contrastive Estimation for Autoencoding-based One-Class Collaborative Filtering [ pdf ]
Jin Peng Zhou, Ga Wu, Zheda Mai, Scott Sanner
Tech Report, 2019
Thesis
Online Continual Learning in Image Classification [ pdf ]
Zheda Mai
University of Toronto MASc Thesis
