MAUSPAD: Mouse-Based Authentication Using Segmentation-Based, Progress-Adjusted DTW

Abstract

The limited transparency of the inner decisionmaking mechanism in deep neural networks (DNN) and other machine learning (ML) models has hindered their application in several domains. In order to tackle this issue, feature attribution methods have been developed to identify the crucial features that heavily influence decisions made by these black box models. However, many feature attribution methods have inherent downsides. For example, one category of feature attribution methods suffers from the artifacts problem, which feeds out-of-distribution masked inputs directly through the classifier that was originally trained on natural data points. Another category of feature attribution method finds explanations by using jointly trained feature selectors and predictors. While avoiding the artifacts problem, this new category suffers from the Encoding Prediction in the Explanation (EPITE) problem, in which the predictor's decisions rely not on the features, but on the masks that selects those features. As a result, the credibility of attribution results is undermined by these downsides. In this research, we introduce the Double-sided Remove and Reconstruct (DoRaR) feature attribution method based on several improvement methods that addresses these issues. By conducting thorough testing on MNIST, CIFAR10 and our own synthetic dataset, we demonstrate that the DoRaR feature attribution method can effectively bypass the above issues and can aid in training a feature selector that outperforms other state-of-the-art feature attribution methods. Our code is available at https://github.com/dxq21/DoRaR.

Classifying Participant Standing and Sitting Postures Using Channel State Information

“…Behavioural biometrics are all non-intrusive [1,30]; however, they are difficult to replicate, which instantly takes away the presentation attacks discussed earlier, and generally more secure in the sense that humans do not have to remember any passwords or token cards. Various behavioural biometrics have been researched, including keystroke dynamics [31][32][33][34], mouse dynamics [35,36], gait [37], and many more.…”

Data driven user authentication with multi-level behavior profiling

“…Rather, we aim to demonstrate that artificiallyinduced motor expertise can complement and enhance the performance of traditional biometric authentication. We do indeed demonstrate this for two existing state-of-the-art mouse based authentication approaches Fu et al (2020); Qin et al (2020). To do so, we design and run a multipleweek IRB-approved experiment on human subjects, and prove that merging artificial intelligence with kinesthetic intelligence provides statistically significant performance benefits.…”

“…These features have a great advantage over histogram features (e.g., in Ahmed and Traore (2007)) as they are movement-based and can verify a user according to individual mouse movements rather than aggregating numerous actions over time. Recently, the authors of Qin et al (2020) proposed a segmentation-based, progress-adjusted DTW algorithm to extract features from the mouse trajectory data.…”

“…Third, the risk scores of the r movements are averaged to produce a final risk score, indicating how likely the current trial is not generated by a legitimate user; this step is called fusion and the final risk score is the output of the classification module. We select these two schemes based on the fact that (i) to the best of our knowledge, and at the time of this writing, they are the most recent works on the topic, showing significant improvements over previous schemes, and (ii) they are fundamentally different and represent two categories of most mainstream mouse-based authentication schemes -one uses explicit feature extraction and a vector-space classifier Qin et al (2020), while the other relies on a deep neural network to implicitly learn from raw data Fu et al (2020). Below, we briefly explain the key ideas of these two schemes.…”