School of
Information Technology and Electrical Engineering

Speaker: Mr Weiting Chen (Tony)
Seminar Date: Wed, 29/11/2017 - 14:30
Venue: 78-631; Data Science MMLab
Host: Prof Xue Li

Seminar Type:  PhD Confirmation Seminar

Abstract: 

Most exiting prediction methods for medical domain have typically been conducted by building one common classification model for all scenarios without considering the correlations between domain features. Learning multiple related tasks simultaneously from correlated data has been proven to be much more effective than learning each of tasks independently. This PhD study focuses on classification tasks on medical data, with the objective of improving the effectiveness and efficiency of the model classifications. We study on multi-task learning algorithms to incorporate the valuable feature correlations in medical data. In specific, we designed new multi-task deep learning models to handle the problems in medical data, such as missing value and inconsistent data. 

Our contributions are in three-fold. First, we develop a novel multi-task deep learning model for classification tasks, including human intention recognition, mortality prediction, and length-of-patient-stay forecasting. Second, we demonstrate that deep learning models can obtain better results on ‘raw’ features without pre-processing. Third, we design a new ICU scoring system that is able to dynamically forecast an ICU patient’s status based on medical examination.

The designed and tested new methods using real medical data sets from EEGMMIDB (EEG Motor Movement/Imagery Dataset), EMOTIV, MIMIC- III (Medical Information Mart for Intensive Care III) and RBWH (Royal Brisbane and Women’s Hospital) databases. Current experimental results show that our proposed Multi-Task Deep Learning method outperforms the state-of-the-art algorithms with a very high classification accuracy in motion intention detection.

Biography: 

Mr. Weitong Chen received his B.S Degree in Information Systems from Griffith Universtity in 2011, and M.S Degree in Computer Science from the University of Queensland in 2013. Currently, he is doing PhD at the School of Information Technology and Electrical Engineering, University of Queensland, under the joint supervision of Prof Xue Li and Prof Michael Sheng (Macquarie University).