Google Play icon

Predicting Hostile Intents using Artificial Neural Networks

Share
Posted January 8, 2015

The problem of identifying suspicious behavior becomes more important almost on a daily basis. We would certainly like to reduce the amount of hostile encounters, but in the world as it is today it would be a good start to create some kind of automated tool which could warn us if something dangerous can potentially happen in the future.

Prediction of hostile behavior is currently limited mostly to the field of military operations. Image credit: U.S. Army via Wikimedia Commons/Public domain

Prediction of hostile behavior is currently limited mostly to the field of military operations. Image credit: U.S. Army via Wikimedia Commons/Public domain

Of course, the science has somewhat limited capabilities in this field as we are still far from creating an artificial intelligence. But existing algorithm-based decision making tools can already provide lots of support, at least on a larger scale.

Let’s consider military tactics. A lot of useful conclusions could be derived simply by analyzing enemy movement patterns. Traditionally, this was done manually (and still is): a person-in-charge looks at a particular chart and tries to make meaningful predictions about movement of armed personnel.

Lately this task is becoming increasingly automated, as decision-making function is being delegated to computers, at least in initial stages of the overall process. For example, recently a team of scientists from India published yet another study describing possibility of using automated computational systems and autonomous algorithms to detect and identify hostile behavior at its early stages.

Screenshot of the actual simulation performed using the new system for predicting hostile behavior. Here, each dot represents an object inside the area of observation. The dot enclosed in the box is controllable by the user. To the extreme right, there is a landmass which is to be protected. Patterns of attack of user-controlled object are used to train the artificial neural network. Image courtesy of the researchers.

Screenshot of the actual simulation performed using the new system for predicting hostile behavior. Here, each dot represents an object inside the area of observation. The dot enclosed in the box is controllable by the user. To the extreme right, there is a landmass which is to be protected. Patterns of attack of user-controlled object are used to train the artificial neural network. Image courtesy of the researchers.

In the field of military this idea is not exactly new, but the range of potential applications is gradually expanding. “The framework posited in this paper has many applications in fields apart from maritime surveillance and radar systems. The system can be deployed in cyber-security domains for analysis and detection of malicious data packets”, the authors note.

The system they proposed is based on artificial neural networks. Inherently, such system has the ability to learn and adapt to new tactics each time a new pattern of behavior is encountered. Currently only locations of objects that could pose a threat are taken as prediction-related factors; however, the authors say that in the future their development could be improved by including other variables into the domain of forecasting.

Certainly, extension of this research into general society (not only to military) is likely not to happen very soon, but let’s hope that this research will promote safety and peace on a global scale.

Written by Alius Noreika

Featured news from related categories:

Technology Org App
Google Play icon
84,754 science & technology articles

Most Popular Articles

  1. Real Artificial Gravity for SpaceX Starship (September 17, 2019)
  2. Top NASA Manager Says the 2024 Moon Landing by Astronauts might not Happen (September 19, 2019)
  3. How social media altered the good parenting ideal (September 4, 2019)
  4. What's the difference between offensive and defensive hand grenades? (September 26, 2019)
  5. Just How Feasible is a Warp Drive? (September 25, 2019)

Follow us

Facebook   Twitter   Pinterest   Tumblr   RSS   Newsletter via Email