Jiliang Tang receives NSF Grant
Jiliang Tang, Assistant Professors of Computer Science and Engineering, has been awarded an NSF grant entitled "Effective Labeled Data Generation via Generative Adversarial Learning".
Recent successes in applying deep learning to solve many challenging data science problems are in part due to the availability of a large number of labeled training data. Reversely, at the same time, lack of labeled training data is still one of the major roadblocks in applying deep learning techniques to challenging data science problems. Existing approaches mainly focus on transfer learning and few-shot learning for alleviating the problem of lacking labeled training data, which requires access to large-scale labeled training data in the source domain. Recent advancement of generative adversarial learning has shown promising results in generating realistic data, which provides a new perspective for alleviating the problem of lacking labeled training data. Hence, in this project, we propose the novel problem of effective labeled data generation via generative adversarial learning, which mainly tackles three challenging tasks: (1) generating labeled data given limited labeled data; (2) generating labeled data with weak supervision; and (3) generating labeled data with human involvement. Within each task, we investigate various challenging subtasks for generating labeled data.
(Date Posted: 2019-08-28)