Skip Ribbon Commands
Skip to main content

Protecting Privacy

​​​​​​

 

With more countries deploying sensors to track the elderly’s health, monitor traffic and accomplish other civic goals, concerns about privacy intrusions have grown proportionately.

To help allay privacy concerns, scientists at Nanyang Technological University’s School of Electrical and Electronic Engineering (EEE), led by Professor Tay Wee Peng have developed an algorithm to enable data collection devices to automatically decide how much and which data to send to centres that analyse the data.

Their work will prevent government agencies and corporations from using collected data to gain information about people that is beyond the stated purpose of the data collection networks.

For example, government agencies may give motion-sensing wearable devicesto the elderly to detect when they fall. The collected data, however, may also reveal information about the elderly’s activities, such as when they go to the toilet or cook.

The EEE team’s algorithm would go through the collected data to determine which data points are necessary for an analysis program to detect a fall, and combine or eliminate the other data points to create a 50 percent error rate for the testing of all other hypotheses.

This means that the analysis program would have enough information to accurately identify whether a fall has occurred, but would be forced to essentially guess whether the data also shows that, for example, the elderly went to the toilet at a particular time.

The EEE scientists validated their algorithm by testing it on data collected to identify when a person sat down or stood up. The data was sent from a body-worn sensor and showed acceleration speeds and gyroscope measurements in different directions.

After the algorithm processed the data, an analysis program identified with high accuracy when the person sat down or stood up but had a nearly 50 percent error rate in guessing whether the person was opening a cabinet.

“Our algorithm can learn by itself, through machine learning, how to process different datasets to achieve the desired outcomes,” said Professor Tay Wee Peng from the EEE.

 


By Professor Tay Wee Peng


Click here to find out more.

Published on: 8-May-2018​​ ​​​
Not sure which programme to go for? Use our programme finder
Loading header/footer ...