Identity authentication based on keystroke latencies using a genetic adaptive neural network
First Claim
1. A system, comprising a processor and a computer-readable medium, the medium being encoded with programming instructions executable by the processor to:
- accept input of a password by a person via a keyboard;
capture keystroke latencies as the person enters the password, wherein the keystroke latencies are only those between characters in a predetermined set of character patterns;
apply the keystroke latencies as input to a neural network that implements a genetic algorithm; and
based on the output of the neural network, generate an authentication signal that relates to whether the person is a particular user.
0 Assignments
0 Petitions
Accused Products
Abstract
A system and method identify the person who is using a keyboard based on keystroke latencies as the person types certain key combinations. In some embodiments the latencies are monitored as the person types a password, while in others they are monitored as the person types other information and continues to use the computer. In some embodiments the identification yields a binary result (whether the latency profile matches the profile stored for a particular user), while in others a confidence level is given. A mismatch, or a confidence level below a particular threshold, results in a request for further identity verification, creation of a log entry, immediate notification of responsible personnel, or denial of access (or continued access).
53 Citations
18 Claims
-
1. A system, comprising a processor and a computer-readable medium, the medium being encoded with programming instructions executable by the processor to:
-
accept input of a password by a person via a keyboard;
capture keystroke latencies as the person enters the password, wherein the keystroke latencies are only those between characters in a predetermined set of character patterns;
apply the keystroke latencies as input to a neural network that implements a genetic algorithm; and
based on the output of the neural network, generate an authentication signal that relates to whether the person is a particular user. - View Dependent Claims (2, 3, 4, 5)
-
-
6. A method of maintaining computer security, comprising:
-
maintaining a collection of latency profiles, each for a particular authenticated user on a computer system;
monitoring the keystroke latencies as a person using the computer system types a predetermined set of n-graphs;
determining a current user as whom the person is logged in;
determining whether the keystroke latencies match the latency profile for the current user; and
if the keystroke latencies do not match the latency profile for the current user, generating an alarm signal. - View Dependent Claims (7, 8, 9, 10)
-
-
11. A system, including:
-
a keystroke latency monitor that records the latency between keystrokes that form one of a predetermined set of n-graphs of keyboard input by a person;
storage that contains a digital keystroke latency signature for a particular user;
a neural network that receives latency data from the monitor, evaluates the latency data against the digital keystroke latency signature, and provides an output signal when the variation between the latency data and the signature exceeds a predetermined threshold. - View Dependent Claims (12, 13, 14, 15, 16, 17, 18)
-
Specification