CHIP: A children identification system based on mobile phone sensory data

Identifiable signatures about people can be left automatically and unobtrusively when they are using smartphones, which can help to identify different users and offer personalized service to them. Specifically, offering services to users according to their age is significant and necessary, as people with different ages have distinct requirements for phone use. However, the smartphone knows little about the user's age. In this paper, we introduce CHIP, a system that achieves CHildren user Identification based on smartphone sensory data only. The system first captures data in two scenes, i.e., unlocking screen and phone answering, by utilizing ubiquitous sensors available in commodity mobile phone. Based on these scenes, we then extract features in the stages of picking up the phone, swiping the screen, response time, and answering the phone. The Random Forest classifier is further used to identify children users. We evaluate the system through real-life experiments conducted by 8396 effective samples from 35 participants with different ages. Experimental results show that our approach achieves the precision of 94.80% in identifying children phone users.