Description
Assignment overview. This assignment is designed to introduce you to a basic neural network. Your task is to modify the MLP program from class that we used to solve the XOR problem to apply it to a letter recognition task.
Questions:
- [20 marks, 15 marks for Grads] Implement a multi-layer perceptron (MLP) by modifying the MLP program from the class to solve the XOR problem and train it to translate the digital letters given in file pattern1 into the corresponding ASCII representation. In the file, each letter of the alphabet is represented by a matrix of 12×13 binary values so that each consecutive 12 rows represent one letter. Plot a training curve and interpret your results.
- [20 marks, 10 marks for Grads] Investigate how much noise the MLP can tolerate in the pattern before being unable to recognize a letter. Explain your implementation of noise and report your results.
- [10 marks, 5 marks for Grads] Which letter is represented in file pattern2?
- Grad Students only [20 marks] Investigate the network performance when training on noisy patterns. Also, how does the number of hidden nodes influence the performance?
`
