چکیده مقاله
Many structures and models are available to design a neural network The accuracy of a neural network depends on its structure, the number of layers, as well as the number of neurons in each layer Therefore, structures and sizes of neural networks are two important factors When there are a number of local criteria in a number of optimization issues In this paper, proposed three layer weight of proposed neural networkwith two W1 and W2 matrices and some neural assumptions in the second and third layers with two vectors are shown In addition, it simultaneously expands the weight of the neural network We continue to optimize the weight of the networks with more layers using the matrix
کلیدواژهها
نویسندگان
Amir Keikhosravi
Student of Engineering, Engineering Department of Electronics Circuits Institute of Higher Education Adiban, Garmsar,Iran
Reza Keneshloo
Student of Engineering, Engineering Department of Electronics Circuits Institute of Higher Education Adiban, Garmsar,Iran
شیوه ارجاع
Keikhosravi, Amir and Keneshloo, Reza,1398,The Optimization of the Structures and Weights of a Neural Network by Means of an Evolutionary Algorithm,Sixth National Congress on Electrical Engineering and Computer Engineering of Iran with a New Approach to New Energy,Tehran
ارائهشده در
مجموعه مقالات ششمین کنگره ملی تازه های مهندسی برق و کامپیوتر ایران با نگاه کاربردی بر انرژی های نو12 اردیبهشت 1398 · تهران