Keystroke Dynamics Authentication Using N-gram Statistical Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing user authentication methods, particularly those based on keystroke dynamics for free text, suffer from loss of dimensionality, leading to reduced accuracy in detecting unauthorized access and potential false labeling of valid users.
Innovation Solution
The method extracts multiple sub-features from each instance of n-grams in typed text, computes statistical features such as standard deviation and mean, and feeds these into a trained machine learning model to enhance authentication accuracy by considering multiple instances and combinations of characters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional password-based authentication is used, then the authentication process is simple and fast, but it is vulnerable to stealing and unauthorized access
Solution Approach 1:
The authentication system segments the authentication process into multiple independent analysis components: n-gram extraction, sub-feature extraction (press-to-release, release-to-press, press-to-press, release-to-release), statistical feature computation, and ML model evaluation. Each component processes specific aspects of keystroke dynamics separately, then combines results for comprehensive authentication decision-making.
Solution Approach 2:
The system transitions from traditional 1D password verification to multi-dimensional keystroke analysis by extracting features across multiple dimensions: temporal dimensions (press-to-release times, release-to-press intervals), spatial dimensions (key positions, typing patterns), and statistical dimensions (mean, standard deviation, skewness). This dimensional expansion creates a much more robust authentication space that is difficult to replicate by attackers.
2Measurement precision
If simple keystroke features are used for authentication, then the processing is fast and simple, but the authentication accuracy is reduced and false positives increase
Solution Approach 1:
The system merges multiple feature extraction approaches into a unified authentication framework. It combines n-gram-based features (capturing sequential patterns), statistical features (capturing temporal variations), and multiple sub-feature types (press-to-release, release-to-press, etc.) into a comprehensive feature vector that is fed into the ML model for accurate authentication decisions.
Solution Approach 2:
The authentication system uses a composite feature structure analogous to composite materials, where multiple types of features (temporal, spatial, statistical) are combined to create a robust authentication signature. Just as composite materials combine different substances to achieve superior properties, this system combines multiple feature types to achieve superior authentication accuracy and resistance to spoofing.
Data Source
AI summary
There is provided a computer implemented method of authenticating a user, comprising: receiving a sequence of key-related events of a manually typed text by a user using a keyboard, extracting a plurality of sub-features from the sequence of key-related events, for each instance of a plurality of instances of a respective n-gram of a plurality of n-grams extracted from the text, computing a plurality of statistical features for each respective n-gram from the plurality of sub-features extracted for the plurality of instances of the respective n-grams, feeding the plurality of statistical features computed for each of the plurality of n-grams into a trained machine learning (ML) model, and triggering a security process when the ML model outputs an indication of non-authentication of the user.