AI Helps Detect Crime and Predict Crime Hot Spots

This post is also available in: עברית (Hebrew)

Law enforcers all over the world are constantly trying to fight crime by detection and prevention, attempting to stay ahead of the criminals. A recent article published in the International Journal of Knowledge-Based Development reports the use of emotional data, machine learning, and deep learning techniques to understand criminal minds and prevent criminal activity by predicting it and acting in advance.

A. Kalai Selvan and N. Sivakumaran, who led the study, had two main objectives: predicting crime using machine learning models based on emotional data and identifying future crime hotspots using deep learning methods applied to crime incident data. According to Techxplore, the researchers analyzed voice-based emotional cues using machine learning algorithms and achieved a detection accuracy of 97.2% for various crimes. They also managed to detect crime hotspots with an accuracy of 95.64% by using deep learning techniques (particularly convolutional stacked bidirectional long short-term memory).

The researchers claim that the emotional states in speech patterns were extremely significant since they allowed them to explore speech-based emotion detection. They reportedly considered linguistic origin, paralinguistic cues, and the characteristics of the speaker, thus integrating the emotional data they obtained with other factors like location and the kind of crime that occurs in a hotspot.

This innovation, while sounding slightly science-fiction-based, today’s rapid advances in algorithms enable to extract and identify patterns in data. The team claims their innovation could be used to monitor activity in crime hotspots, detect crimes, and even forecast future criminal activities.

Further development and future work could potentially apply similar machine-learning techniques for emergency response systems. For example – by analyzing the emotions of a person calling the emergency services, the system might be able to distinguish between real emergencies and non-emergency or even prank calls. This could significantly reduce the burden on the services and enable quicker assistance to those in need.