A great real-world use case to demonstrate the power of MLPs is that of OCR. In OCR, the challenge is to recognize human writing, classifying each handwritten symbol as a letter. In the case of the English alphabet, there are 26 letters. Therefore, when applied to the English language, OCR is actually a classification problem that has k = 26 possible classes!
The dataset that we will be using has been derived from the University of California's (UCI) Machine Learning Repository, which is found at https://archive.ics.uci.edu/ml/index.php. The specific letter recognition dataset that we will use, available from both the GitHub repository accompanying this book and from https://archive.ics.uci.edu/ml/datasets/letter+recognition ...
