Researchers develop accurate gender identification technique with keystroke dynamics
New research shows that keystroke dynamics can be used to determine the gender of a person typing with accuracy as high as 95.6 percent, IFLScience reports.
“Keystroke dynamics features for gender recognition” was recently published in the journal Digital Investigation by a trio of researchers from the Department of Electrical and Computer Engineering, School of Engineering, Democritus University of Thrace in Greece. They tested the accuracy of five different models on a variety of datasets, and created a new dataset by recording 75 individuals’ daily computer usage over a 10-month period with a program called “ISqueezeU” to determine the typing characteristics most suitable to gender recognition, according to IFLScience.
While “ISqueezeU” identified user gender with 78 percent accuracy, it also revealed to the researchers that the amount of time between pressing the “N” and “O” keys is the strongest keystroke dynamic feature for identifying gender.
The researchers achieved over 95 percent accuracy, which they say is the highest found in related literature, by applying a few hundred features.
As previously reported, TypingDNA launched a keystroke dynamics authentication solution in December based on short text samples.