Over the past decade, deep learning has revolutionized computer vision, multimedia, natural language processing, and all other artificial intelligence sub-fields. Nowadays, these well-trained deep learning models can significantly outperform our humans in many computer vision tasks, including complex visual scene understanding (e.g., visual-language tasks). Despite unprecedented attention and great success, today’s visual scene understanding models still fail to realize human-like understanding. By “human-like”, we mean that these vision systems should be equipped with three types of abilities: 1) Explainable: The model should rely on (right) explicit evidences when making decisions. 2) Robust: The model should be robust to some situations with only “low-quality” training data. 3) Universal: The model design is relatively universal, and it is expected to be effective for various tasks or architectures.
In this talk, the speaker will show his previous contributions in these directions and introduce some ongoing and future plans towards next-generation human-like AI and computer vision systems.
About the Speaker
Dr Long Chen is currently a postdoctoral research scientist at the School of Engineering and Applied Science, Columbia University, USA. He obtained his Ph.D. degree in Computer Science from Zhejiang University in 2020. His primary research interests are computer vision, multimedia, and machine learning. His research work has won several awards and honors, e.g., the 2020 Best Ph.D. Thesis of Zhejiang University and Zhejiang Province, the 2021 Global Top-100 Chinese Rising Stars in AI, the Best Paper Award of the HUMA workshop in ACM MM 2021. His team has won the first place in the International Video Relation Understanding Grand Challenge in 2021.
He is a tenure-track faculty candidate for the Artificial Intelligence & Data Science, Machine Learning & Intelligence cluster.
