Updated weights of CNN

3 views (last 30 days)
Md. Rokibul Islam
Md. Rokibul Islam on 14 Jan 2018
Answered: Ronit on 20 Sep 2024
Dear friends, I can initialize weight and view updated weight of Convolutional NN. i need weights of individual iteration/ epoch. is there anyone to help me in this purpose.
Thank you in advance. Rokibul

Answers (1)

Ronit
Ronit on 20 Sep 2024
Hello Rokibul,
There are a few ways which can be used to save the weights of a Convolutional Neural Network (CNN) at each iteration or epoch.
  1. “getwb” function – This MATLAB function can be utilised if working with Neural Network Toolbox. It retrieves the weights and biases of a network as a single vector. Please use the documentation link for further details: https://www.mathworks.com/help/deeplearning/ref/getwb.html
  2. For Convolutional Neural Networks (CNNs) created using the Deep Learning Toolbox, the weights can be accessed through the “Layers” property of the network. Firstly, access a specific layer using indexing, and then directly access the “Weights”. If you want to extract “weights” from all layers, a “for loop” can be used.
% Assume 'net' is your trained network
% Access the first convolutional layer
convLayer = net.Layers(2); % Adjust the index to match your network
% Extract weights
weights = convLayer.Weights;
I hope it helps your query!

Categories

Find more on Recognition, Object Detection, and Semantic Segmentation in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!