Наредни састанак Семинара биће одржан онлајн у среду, 13.априла 2022, од 19-20 часова.
Предавачи: Piotr A. Kowalski, Faculty of Physics and Applied Computer Science, AGH University of Science and Technology; Systems Research Institute, Polish Academy of Sciences
Наслов предавања: SENSITIVITY ANALYSIS FOR ARTIFICIAL NEURAL NETWORK STRUCTURE REDUCTION
Апстракт: In the seminar, I will show some propositions of the use of sensitivity analysis (SA) for the structure reduction of the selected neural network topologies. Two algorithms are introduced. The first algorithm applies SA for the neural network input layer reduction by selecting significant features of input patterns. The second algorithm utilizes the SA procedure to remove redundant neurons of the network. The classification qualities of reduced and full structure neural networks are compared. Furthermore, we set for comparison the neural networks for which different state-of-the-art reduction methods are applied both in the input layer and pattern layer. The models are tested in the classification problems of some repository data sets.
Напомена: Регистрациона форма за учешће је доступна на линку:
https://miteam.mi.sanu.ac.rs/asset/CW5nJWDSEZDj7p32p
Уколико желите само да пратите предавање без могућности активног учешћа, пренос је доступан на линку:
https://miteam.mi.sanu.ac.rs/asset/4LNW8WtML7rLKojoz