School of Information Technology - Page 11

A QUICK GLANCE AT WHAT WE ’ RE WORKING ON …

PROTECTING PERSONAL DATA
We ’ re working with Yonsei University in South Korea and the Electronics and Telecommunications Research Institute to create Bio-PIN , a secure set of algorithms that can be incorporated into any biometric cryptosystem . The Bio-PIN has an extra layer of encryption that protects the raw biometric data . With biometric data playing an increasingly important role throughout society , this is an important initiative in ensuring personal security .
PREDICTING WORSENING HEART FAILURE
We ’ re working on developing a ‘ small data ’ machine learning algorithm in an application , which can predict early worsening outcomes in heart failure patients . The application applies intelligent software agent system design and methods installed on a low-cost consumer smartphone , biosensor or wearable , allowing for routine patient management that will help hospitals remotely monitor their patients and enhance the quality of care with proactive medical therapy .
MAXIMISING ENERGY EFFICIENCY
The expanding use of ICT globally is driving an alarming increase in energy consumption and its carbon footprint . As the world moves to adopt 5G-enabling technologies , it ’ s targeted that future cellular communications must pull 30 % of power from renewable energy by 2035 . To support that vision , we ’ re working on energy sustainable paradigms for 5G and Beyond 5G networks to maximise the efficiency of the harvested energy while minimising grid power consumption .
ASSISTING CANCER DIAGNOSES
We ’ re writing an algorithm to assist medical experts in the timely diagnosis of breast cancer . The deep learning algorithm uses a convolutional neural network to sort mammography images into malignant and benign categories without prior information on the presence of a cancerous lesion . Our researchers have also been successful in creating a segmentation method that can accurately identify a brain lesion and its location from CT images of stroke patients .
STOPPING CYBERBULLIES
Under the anonymity of the internet and the growing popularity of social media , cyberbullying continues to be a serious threat to the mental health and wellbeing of its victims . We ’ re adapting emotion mining techniques to detect instances of cyberbullying , taking into account culturally-bound words and dialects for Malaysian contexts . This will help administrators to identify cyberbullies and suspend malicious attacks .
DETECTING CRIMINAL ACTIVITY
The conventional video surveillance for urban security is not necessarily an effective crime prevention tool . Most of the time , the recorded content is used primarily as evidence of a crime that has already taken place . Our researchers are designing an active surveillance system that can analyse the content of the video feed in real-time to identify potential suspicious behaviours that may lead to criminal intent .
9