Lie will soon Detected on Computer Programs

Lie will soon Detected on Computer Programs - Inspired by the work of psychologists who study the human face as a guide a person has lied, computer scientists Ifeoma Nwogu, Nisha Bhaskaran and Venu Govindaraju examine whether the machine can also read the visual cues that are deceptive.

The results are promising so far. In a study involving 40 conversations recorded by video tape, an automated system that analyzes the eye movements were able to identify exactly what the interview subject is lying or telling the truth as much as 82.5 percent.

"The level of accuracy is better than an interrogator who is usually on account of an experiment to detect lies," said Ifeoma Nwogu, an assistant professor in UB's Center for Unified Biometrics and Sensors (CUBS), which helped develop the system.

The result, even though only an experienced interrogator who has an average of 65 percent closer to the truth.

"What we want to understand is whether there are changes in the signals emitted by people when lying and the machine can detect it? The answer is yes and yes," he said as quoted by Medindia Nwogu, Wednesday (04/04/2012).

After a lie-detector technology with a hot body and facial expressions, UB automated system to track the different characteristics of eye movements.

The system uses statistical techniques to create a model of how people move his eyes in two different situations.

During normal conversation, and when answering questions designed to encourage a lie.

People who changed his pattern of movement between the first and second scenarios are assumed to lie, while the people who have a consistent eye motion is assumed to tell the truth.

What Nwogu and his team are doing is creating an automated system that can verify and correct coding information used by humans in order to successfully classify liars and dishonest people.

The next step is to expand the number of subjects studied and developed automated systems that analyze body language in addition to contact mata.Nwogu said that although the sample slightly, but the results are quite impressive.

Researchers also stated that the computer may be able to study human behavior in the short term to help the task is quite heavy, even for a seasoned interrogator.

"But this does not mean that a machine is ready to replace the human questionnaire, only a computer can be a useful tool to identify a lie," said Nwogu.

Researchers used 40 video recordings of conversations to establish a normal eye movements and fundamental to every subject, especially the number and frequency flicker in which the subject changed the direction of gaze.

Then the researchers used an automated system is to compare the eye movements on the basis of each subject with a critical eye movements during the session of each interrogation, the interrogator to stop asking the point where ordinary things and began to ask to check the truth of his words.

If the machine detect the variation of the basic eye movements are not uncommon, researchers predict that the subject is lying.