THE MULTI-STAGE TRAINING METHOD FOR A CONVOLUTIONAL NEURAL NETWORK
Journal Title: Международный научный журнал "Интернаука" - Year 2018, Vol 1, Issue 6
Abstract
This paper describes methods and algorithms for training neural networks. Also, the organization of multi-stage training of the neural network, based on adaptive and genetic methods, which has been successfully applied to the convergent neural network to solve the problem of object classification.
Authors and Affiliations
Oxana Tarasenko-Klyаtchenkо, Viktoriia Buts
Keywords
Related Articles
- EP ID EP599396
- DOI -
- Views 73
- Downloads 0
PREDICTING LABOR MARKET DOWNTURNS IN UKRAINE USING MARKOV APPROACH
The possibility of solving the crisis prediction problem based on hidden Markov models was investigated. An early warning system for labor market crises in Ukraine was created.
PROTECTIVE HELMET FOR EXTREME CONDITIONS
The article presents the results of using the binary system cooling rescuer safety helmet. It is found that this allows a long time to carry out work under the influence of the presence of excess radiation protection fro...
STUDY OF PHASE TRANSITIONS OF CELLULOSE NANOCRYSTALLITES
In this paper phase states and phase transitions of nanoscale cellulose crystallites have been studied. It was found that conditions of the phase transitions of rod-like nanocrystallites are depended on their lateral siz...
DEVELOPMENT OF METHOD OF COMPLEX QUANTITATIVE ESTIMATION OF QUALITY OF SAUCE MILK
The hierarchical structure of the complex indicator of the quality of the dairy sauce by the method of qualimetry.
MAIN TRENDS OF DEVELOPMENT OF THE CREDIT MARKET OF UKRAINE
The article deals with the main problems and trends of the development of the credit market of Ukraine in the conditions of financial and economic instability. The main directions of ensuring the effective functioning of...