Crime & Machine Learning

In recent years, the development of machine learning algorithms has led to a revolution in law enforcement. Incorporating machine learning into policing gives us a better intuition of the crime committed. We will dive into how ML is used in the fight against criminals, and the potential benefits and challenges associated with this technology.

The most commonly used machine learning algorithms are predictive policing algorithms. Predictive policing tools are built by feeding data — such as crime reports, arrest records, and license plate images — to an algorithm, which is trained to look for patterns to predict where and when a specific type of crime will occur. This information can then be used to allocate resources more effectively and prevent crime before it occurs. For example, law enforcement agencies can use predictive policing algorithms to identify high-crime areas and deploy extra patrols or other measures to prevent crime from taking place. Proponents argue that predictive policing can help predict crimes more accurately and effectively than traditional police methods.

Crime analysis is another important application of machine learning. Machine learning algorithms can analyze large amounts of crime data, identify trends and patterns, and provide insights into the motivations and methods of criminals. This helps officers make well-informed decisions.

Facial recognition has been responsible for innumerable arrests. Facial recognition searches that lead to criminal charges most commonly begin with an image, often from security cameras. That photo is run through a system that compares the image to those in a large database, like a collection of mugshots or driver’s license photos. If the match is found it can be admissible in court as evidence.

In the 19th century, Dr. Cesare Lombroso claimed that people’s physical features reflected their moral character. Many researchers have tried to come up with an application that can differentiate between a criminal and a non-criminal. In recent times, there’s been a study by Dr. Xiaolin Wu and Dr. Xi Zhang that used an AI to make the distinction by analyzing pictures of law-abiding citizens and criminals. They claim that “unlike an examiner/judge a computer vision algorithm or classifier comes without subjective baggage, having no emotion, no biases whatsoever due to past experiences”.

Figure 1. (a) FGM results; (b) Three discriminative features ρ, d, and θ. ”Criminals have a smaller angle theta, and larger curvature rho”.

Figure 2. (a) and (b) are ”average” faces for criminals and noncriminals generated by averaging of eigenface representations ; (c ) and (d) are ”average” faces for criminals and non-criminals generated by averaging of landmark points and image warping.

Despite the many benefits of machine learning in law enforcement, there are also some potential challenges associated with this technology. One of the main concerns is the bias in the training data used to train the various algorithms. This can result in unequal treatment of different groups, and perpetuate existing biases in the criminal justice system. Facial recognition systems are more likely to misidentify people who are not white men, including people with dark skin, women, and young people. There’s a considerable spectrum of accuracy and image quality remains an issue. NIST’s most recent test, which largely relies on a database of high-quality mugshot photos, found that even the best algorithms can be wrong more than 20 percent of the time.

In conclusion, there are many potential challenges with this technology, as with any new technology, which need to be addressed to ensure the effective use of same. With each passing day, better algorithms are being created and technology is being refined. Maybe one day we’ll be able to identify a criminal using machine learning, till then the sky is our limit.