Anomaly Detection
Detection of anomalous situations from the modeling of house habits, aiming at the residents' safety.
What is it? The aim of this work is to build a smart home monitoring system that is able to map the data received from IoT devices in daily activities, such as eating, sleeping and cooking, and to evaluate these activities as normal or abnormal.
Why does it matter? Usual techniques and methods of residential security have something in common: they are devices that aim to actively block or discourage the criminal from committing the crime. However, when perceiving the level of security, the criminal who wants to commit the crime seeks solutions to circumvent the system more and more.
What are the results? Our system uses two different approaches to detect patterns: Random Forest through supervised learning, and a time series algorithm, the latter being an unsupervised approach.
USP Polytechnic School Computer Engineering
Capstone Project 2019
Students: Hugo Su and Victor Barreira
Advisor: Reginaldo Arakaki
Co-advisor: Victor Hayashi
