Predictive Analytics In The Fight Against Terrorism

Is it right and should we be using it?


The Paris and San Bernardino massacres have left the world gripped by fear of further terrorist attacks. The aftermath has already seen planes grounded, schools closed, and newspapers filled with Henny Penny, ‘Sky Is Falling’ headlines claiming the impending Armageddon.

The arguments for combating the spread of Islamo-Facist groups are many and persuasive, while those against are usually founded on cheap relativization along the lines of ‘they were provoked by inflammatory cartoons’. There is now universal approval - in the West at least - of something being done. The question is simply what this ‘something’ is, and, as the UK debate about whether or not to launch air strikes proved, there are many points of view to consider.

Defending against attacks at home has also raised questions. Increased surveillance is the obvious answer, but many fear that doing so will bring about a Police State. Invasion of privacy by the government, or whoever they outsource it to, is an issue, although it is something most are prepared to sacrifice - rightly or wrongly. There are two central problems with increased surveillance. Firstly, that governments use it to identify people that they believe simply to be subversives, those who pose no existential threat to the population, only to the government’s hold on power. This is not so much a problem with invasion of privacy, as it is a lack of trust in government. Secondly, there is the issue that surveillance could identify the wrong people, often through racial profiling. This is a problem with the inability of the people to accurately analyze the surveillance, and the application of predictive analytics to surveillance technologies could increasingly see this cease to be a problem.

There are a number of ways in which predictive analytics can help drive better surveillance. The development of video content analysis (VCA) software is in its early stages, but new algorithms for image analysis, especially facial recognition systems, are getting much better at identifying human faces, particularly in a crowd. Predictive analytics can also look at video surveillance to identify potentially threatening human behavior in public spaces by analyzing patterns found in video footage of similar past events. New software is being developed that builds models for fully-automated semantic-tagging of surveillance video recordings based on multiple human presence detection and abnormal activities recognition.

However, there is a danger in turning surveillance over to technology. These algorithms could end up reflecting biases inherent in the data they are searching, and cause problems especially with racial profiling. If race is disproportionately (but not explicitly) represented in the data fed to a data-mining algorithm, it is possible that the algorithm can infer race and use it indirectly to make an ultimate decision. The attacks on the West have been made by Islamic terrorists, and the video footage under analysis for terrorist attacks will show perpetrators of middle eastern origin. The pitfalls in this are that firstly that racial profiling will occur and the wrong people potentially penalized, and, secondly, that potential white attackers will be missed.

Friedrich Nietzsche perceived that Western civilization was moving in the direction of the Last Man, an apathetic creature lacking both great passion or commitment. 'Unable to dream, tired of life, he takes no risks, seeking only comfort and security, an expression of tolerance with one another.’ It is important to remember, however, that there are many risks in seeking comfort and security. We will have to judge whether these are risks worth taking, and whether predictive analytics is the right tool to use.


comments powered byDisqus
Bottom of well small

Read next:

Data Mining In The Deep Web