Mr Alex Peng
Research | Current
The goal of my PhD research is to improve the prediction accuracy of neural networks with unlabelled data. Neural networks are a type of machine learning models that are widely used today in tasks such as machine vision, natural language processing and speech recognition. To achieve good performances on these tasks, a neural network has to be trained on a large amount of labelled training data (e.g. an image tagged with “cat”). However, labelled data can be difficult and expensive to collect in some applications. Fortunately, unlabelled data (e.g. an image without tags) is usually readily available. In my research, I study different strategies of using unlabelled data to improve the performance of neural networks when labelled data is scarce.
Selected publications and creative works (Research Outputs)
- Peng, A. Y., Sing Koh, Y., Riddle, P., & Pfahringer, B. (2019). Using supervised pretraining to improve generalization of neural networks on binary classification problems. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 10.1007/978-3-030-10925-7_25
Other University of Auckland co-authors: Patricia Riddle, Yun Sing Koh
- Peng, A. Y., Koh, Y. S., & Riddle, P. (2017). mHUIMiner: A fast high utility itemset mining algorithm for sparse datasets. In J. Kim, K. Shim, L. Cao, J. G. Lee, X. Lin, Y. S. Moon (Eds.) Advances in Knowledge Discovery and Data Mining: 21st Pacific-Asia Conference, PAKDD 2017 (Lecture Notes in Computer Science), 10235 (Part 2), 196-207. Jeju, South Korea. 10.1007/978-3-319-57529-2_16
Other University of Auckland co-authors: Yun Sing Koh, Patricia Riddle