Lab1: Backpropagation

Important Date

Notice: zip all files in one file and name it like「DLP_LAB1_your studentID_name.zip」

Implement simple neural networks with two hidden layers.
You must use backpropagation in this neural network and can only use Numpy and other python standard libraries to implement.
Plot your comparison figure that show the predicted results and the ground-truth.

Introduction (20%)
Experiment setups (30%)
- Sigmoid functions
- Neural network
- Backpropagation
Results of your testing (20%)
- Screenshot and comparison figure
- Show the accuracy of your prediction
- Learning curve (loss, epoch curve)
- Anything you want to present
Discussion (30%)
- Try different learning rates
- Try different numbers of hidden units
- Try without activation functions
- Anything you want to share
Extra (10%)
- Implement different optimizers (2)
- Implement different activation functions (3)
- Implement convolutional layers (5)

60% demos core (experimental results & questions)
40% report
For experimental results, you have to achieve at least 90% of accuracy to get the demo score.
If the zip file name or the report spec have format error, you will be punished. (-5)