Unsupervised learning-based detection method and driver profile- based vehicle theft detection device and method using same

Abstract

An unsupervised learning-based detection method according to one technical aspect of the present disclosure relates to an unsupervised learning-based detection method using a supervised-learned model, and includes: generating a first plurality of matrix data on the basis of driving data; generating encoding information by encoding the first plurality of matrix data using a convolutional neural network; modeling a time series feature of the encoding information by using a long short-term memory (LSTM) network, so as to derive a correlation between variables according to a time series; re-implementing a second plurality of matrix data through a deconvolution calculation of the correlation between the variables according to the time series; and determining whether the driving data corresponds to a pre-supervised learned driver profile, on the basis of a difference between the first plurality of matrix data and the second plurality of matrix data.