---
titles: HW9
tags: teach:MF
---

# HW9

In this homework assignment, you will download the MNIST DATABASE of handwritten digits in the green box below on website http://yann.lecun.com/exdb/mnist/ to solve the following problems.

(Hint: Try to download the dataset directly in the program with code will be easier. )


The explanation of database as follow：
The MNIST database of handwritten digits, has a training set of 60,000 examples, and a test set of 10,000 examples. It is a subset of a larger set available from NIST. The digits have been size-normalized and centered in a fixed-size image.


Analysis the data.
PLZ Mark The Question Number Clearly!

(Hint: For doing this homework, you can use “Google Colaboratory” for free GPU to overcome the difficulties of computer hardware if you need it !)

1. Try to print out 0~9 so that knowing the things you are going to input in the model.
2. Using CNN for handwriting recognition. PLZ give the reason why you choose the specific Loss function, Optimizer, etc.
3. Try to print out some output to make sure it is right.
4. Choose one model you had learned in previous lecture, using the model you choose for handwriting recognition.
5. Compare the result in CNN and the model you choose, which one has the better performance ? (Hint: you can select a index such as accuracy, ROC, recall, etc to compare the performance.)