TypeNet: Scaling up Keystroke Biometrics

Abstract

We study the suitability of keystroke dynamics to authenticate 100K users typing free-text. For this, we first analyze to what extent our method based on a Siamese Recurrent Neural Network (RNN) is able to authenticate users when the amount of data per user is scarce, a common scenario in free-text keystroke authentication. With 1K users for testing the network, a population size comparable to previous works, TypeNet obtains an equal error rate of 4.8% using only 5 enrollment sequences and 1 test sequence per user with 50 keystrokes per sequence. Using the same amount of data per user, as the number of test users is scaled up to 100K, the performance in comparison to 1K decays relatively by less than 5%, demonstrating the potential of Type-Net to scale well at large scale number of users. Our experiments are conducted with the Aalto University keystroke database. To the best of our knowledge, this is the largest free-text keystroke database captured with more than 136M keystrokes from 168K users.

SPCAuth: Scalable and Privacy-Preserving Continuous Authentication for Web Applications

“…All three approaches provide good accuracy, but only scale to a limited extent as the number of users increases, because the models are designed to recognize a single user and need to be retrained, or additional models need to be established when new users join. TypeNet [6], on the other hand, determines from a sequence of keystrokes whether they originate from the same user or if an imposter is involved. Moreover, Acien et.…”

“…16KIS1142K incoming requests upon arrival at the application server and check whether transmitted information, such as IP addresses, are consistent with previous requests of the same user [4], [5]. Additionally, it can be helpful to observe user interactions with the application and detect discrepancies, for example by observing typing patterns, touchscreen usage and the physical location of the devices used [6]- [9]. As a result, continuous authentication settings [10] can be established, where the user's behavior is permanently evaluated and authentication decisions are revised.…”

SetMargin Loss applied to Deep Keystroke Biometrics with Circle Packing Interpretation

“…Partially Observable Hidden Markov Models were employed in [6] for free-text keystroke verification obtaining a competitive accuracy. More recently, the availability of large scale databases with millions of keystroke samples has allowed training deep models with very competitive performances in free-text scenarios [7]. The architecture proposed in [7], called TypeNet, was trained using a Contrastive Loss function with performances six times better than previous approaches based on traditional statistical methods [6,8].…”

“…More recently, the availability of large scale databases with millions of keystroke samples has allowed training deep models with very competitive performances in free-text scenarios [7]. The architecture proposed in [7], called TypeNet, was trained using a Contrastive Loss function with performances six times better than previous approaches based on traditional statistical methods [6,8]. Our purpose in the present paper is to improve further the state-of-the-art results of deep keystroke biometrics by introducing a new loss function expected to be also useful in other challenging recognition problems.…”

Assessing the Quality of Swipe Interactions for Mobile Biometric Systems

“…In particular, some datasets have been collected in the last years including behavioural biometrics (e.g. keystroke, swipe, signature) [1,2,3,4]. In particular, swipe biometrics is focused on authenticating the user, continuously or on queries, based on behavioural informations gathered from fingers interaction 978-1-7281-9186-7/20/$31.00 c 2020 IEEE with a device touch screen.…”