The Use of Artificial Neural Networks in Forensic Science: Actual Problems and Solutions

 The development of machine learning methods has led to the proliferation of artificial neural networks, the architecture and basic principles of which conditionally simulate the process of biological neural networks, mainly the human brain. The idea of such modeling arose as a result of the desire to artificially reproduce some of the qualities of biological neural networks that allow a person to accumulate experience and solve emerging problems based on it. Such qualities implemented in artificial neural networks are the ability to learn and correct mistakes. In the field of forensic science, there are already cases of using neural networks to solve expert problems. At the same time, a forensic examination can only be conducted on the basis of scientifically sound data, allowing to verify and evaluate the reliabi­lity of the conclusions reached. During conducting an examination, the expert should be guided by the methodological recommendations developed for this type of expertise. In turn, artificial neural networks have a number of features that make it difficult and sometimes impossible to use them in the process of forensic examination.

The purpose of the research is to analyze the functioning features of artificial neural networks in the context of their using in forensic science and identify the most pressing problems in this field.

The methodological basis of the research was the dialectical method of cognition along with the systematic approach, general scientific methods: deduction and induction, analysis and synthesis, the logical method, as well as private scientific research methods: formal-logical, systemic-structural, logical-legal.

Currently, the using of artificial neural networks in forensic science is possible only as the experimental developments or outside the process of forensic examination when checking objects according to forensic records. The primary problem hindering the practical implementation of artificial neural networks in forensic science is the lack of scientific and methodological justification. The development of methodological recommendations should begin within the framework of the general theory of forensic science, which in the future may become the basis for the development of methods for certain kinds (types) of forensic examinations. In general, an artificial neural network can be recognized as only one of the ways to solve an expert problem. The determining role in the formulation of the final conclusion should remain with the expert.