Premium
Lie detection using extreme learning machine: A concealed information test based on short‐time Fourier transform and binary bat optimization using a novel fitness function
Author(s) -
Dodia Shubham,
Edla Damodar R.,
Bablani Annushree,
Cheruku Ramalingaswamy
Publication year - 2020
Publication title -
computational intelligence
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.353
H-Index - 52
eISSN - 1467-8640
pISSN - 0824-7935
DOI - 10.1111/coin.12256
Subject(s) - lie detection , computer science , artificial intelligence , pattern recognition (psychology) , extreme learning machine , binary classification , speech recognition , curse of dimensionality , brain–computer interface , support vector machine , machine learning , electroencephalography , artificial neural network , deception , psychology , social psychology , psychiatry
Lie detection is one of the major challenges that is being faced by the forensic sciences. Identification of lie on the basis of a person's mental behavior is a tedious task. Brain‐computer interface is one such medium which provides a solution to this problem by displaying visual stimuli and recording subject's brain responses. A P300 response is elicited whenever a person comes across a familiar stimuli in a series of rare stimuli. This P300 response is used for the lie detection method. In the proposed concealed information test, acquired signals are preprocessed to discard noise. Then, short‐time Fourier transform method is applied to extract features from the preprocessed electroencephalogram signals. To avoid the curse of dimensionality and to reduce computational overhead, binary bat algorithm is applied, which helps in choosing optimal subset of features. The obtained set of features is given as an input to the extreme learning machine classifier for training of guilty and innocent samples. The performance of the system is assessed using 10‐fold cross‐validation. The resultant accuracy obtained from the proposed lie detection system is 88.3%. The system has provided efficient results in contrast with most of the state‐of‐the‐art lie detection methods.