Keystroke dynamics is a technique in which unique patterns from the timing between pressing and releasing of keys during the keyboard usage are recorded for carrying out authentication. This field has been emerged drastically from past two to three years with advancements in artificial intelligence (AI). It can be used as a part in a multifactor authentication system for implementation of various biometric verification techniques. Various research studies have been published that outlines how the keystroke dynamics are becoming influential for various applications. According to the report published by Allied Market Research, the global keystroke dynamics market is expected to generate $754.86 million by 2025, growing at a CAGR of 24.7% from 2018 to 2025. Following are some of the research activities that outline the changing scenario of the market:
Determining gender of a person using keystroke dynamics:
The recent research activities have provided interesting findings related to keystroke dynamics. According to the report by IFLScience, keystroke dynamics is useful for determining the gender of person who is typing with nearly 96% accuracy. In a report titled, Keystroke dynamics features for gender recognition, published in Digital Investigation journal, the findings from the program “ISqueezeU” have been published. The accuracy of five models was tested on various datasets and new dataset was created by noting down the computer usage of 75 individuals for a period of 10 months. The program outlined the interesting findings. It showed that the amount of time taken for pressing “N” and “O” was the prime factor in identifying gender. After application of few hundred features, researchers were able to gain nearly 95% of accuracy. They claim that it is the highest accuracy that has been achieved in the related studies.
Advancements in machine learning to transform keystroke dynamics:
According to Raul Popa, CEO and data scientist of TypingDNA, stated that many investors have invested an enormous amount of money into improvement of accuracy of typing biometrics. However, advancements in machine learning has resulted in the true success over the past couple of years. TypingDNA has utilized the advancements for development of AI-powered typing pattern recognition technology. The company claims that it is more than 99% accurate. It stated that the technology would reach nearly 99.9% accuracy if the sufficient amount of typing profile available. Various dimensions have been used, such as the way users type, time taken to move from one key to another, and the time for which each key is pressed. These dimensions create a unique pattern that can be used to represent feature vectors. The technology of the company records the data related to 44 most used keys of the keyboard. It needs a user to type minimum 60–70 characters. However, there are different requirements based on requirements, the error threshold can be adjusted based on the needs of an owner. Then the balance between accuracy and usability can be achieved. The system of TypingDNA can determine changes in typing style. If it changes too much, the configured accuracy threshold determines the success and failure of authentication.