On the Question of the Applicability of Artificial Intelligence Systems to Forensic Examination of Documents and Their Requisites

Abstract

The article describes the possibilities of applying artificial intelligence systems used in a number of tasks arising in the course of law enforcement activities and also beyond them. The author gives examples of solving identification, prediction, and detection problems that are successfully solved using such systems. Based on the analysis of the principles of the construction and functioning of the neural network, which forms the basis of artificial intelligence, as an abstract model created to solve a specific problem, the capabilities of neural networks as a whole are revealed, and thus the features of working with them are demonstrated. The procedure and principles of operation of convolutional neural networks are described and, using their example, the results that can be achieved during forensic handwriting and technical forensic examinations of documents, as well as studies conducted within the framework of forensic expert activities, in particular, the identification of signs of technical forgery of documents (installation, erasure, additions, finishing) are indicated, establishing the facts of the execution of several signatures or handwritten records by one or different performers, establishing the original content of the records, fragments of which have been partially lost or distorted. The prospects and significance of convolutional neural networks are revealed, the use of which in the expert research process makes it possible to increase the scientific validity and objectivity of the conclusions of forensic examination and, thereby, creates prerequisites for increasing its role in proving facts of importance.