lab1

Back-propagation

Zhi-Yi Chin, National Yang Ming Chiao Tung University

In this lab, we will need to understand and implement simple neural networks with forwarding pass and backward propagation using only two hidden layers. The simple neural network we are going to implement is a feedforward neural network.

Setup (with Anaconda)

Install dependencies: conda env create -f environment.yml

Activate environment: conda activate backprop

Datasets

1. Linear data: ./data/linear_data.csv
2. XOR data: ./data/xor_data.csv

Pretrained Models

1. Linear case: ./checkpoints/linear_0.01_(512,32).pkl
2. XOR case: ./checkpoints/xor_0.1_(512,32).pkl

Training and Testing

python main.py [-h] [--task TASK] [--lr LR] [--hidden-size HIDDEN_SIZE] [--load LOAD]

optional arguments:
  -h, --help            show this help message and exit
  --task TASK           which kind of data to process: [linear, xor]
  --lr LR               learning rate
  --hidden-size HIDDEN_SIZE
                        two hidden layer neuron numbers (layer1,layer2)
  --load LOAD           if testing weight is loading from file

ex:

python main.py --task xor --lr 0.1 --hidden-size (512,32) --load False