Intelligent systems. Task for the laboratory for hand written character recognition
Practically test the application of specialized artificial neural network MATLAB (or alternative) functions to solve practical problems. Learn to quickly test multiple different neural networks for the same task.
Recognize the handwritten numbers in the image using your own created or specialized MATLAB functions. For image processing you can take advantage of the already developed feature extractors (image segmentation, uniform, 50x70 pixels, preparation of size squares with numbers, 35 character extraction ...). You need to prepare your images with handwritten symbols by your own. You can use the prepared "vaizdo_atpazinimas.m" file, which uses the "pozymiai_raidems_atpazinti.m" function for character recognition and training (you need to specify the image name and the number of character strings in the image). Decrease the number of RBFNs for recognizing neurons from 13 to any selected smaller number and check how well the network succeeds in recognizing your handwritten words or numbers.
Solve the same problem using a multilayer perceptron (e.g., using "newff" or "feedforwardnet" functions).