Projects & Grants




Machine learning as a tool for sequence processing.
Project IdSGS13/PřF/2023
Main solverIng. Petr Dvořáček
Period1/2023 - 12/2023
ProviderSpecifický VŠ výzkum
Statesolved
AnotationIn this project, we deal with current machine learning methods that are specialized in sequence processing. We will focus on the shortcomings of these methods and propose solutions to overcome their shortcomings in our own research. Namely, we will extend existing methods for sequence processing (classification, future prediction, pattern detection, compression) to include the possibility of processing multispectral sequences; streamlining the methods used in reinforcement learning; and addressing the issue of neural network security. The implications of our research are at the application level, where our solutions can be implemented in applications using big-data or reinforcement learning. The output of neural network security research is also applicable in critical tasks that are controlled by neural networks, which is, for example, typical for the implementation of robots in industry.