Title: Performance comparison and analysis on Perceptron with different activation functions and loss functions.

Building two Perceptrons with the following two groups of activation functions and loss functions on the training set $($train$\_$catvnoncat.h$5)$

Group $1$:

Activation function:

$a_{w,b}(x^i)=\frac{1}{1+e^{-(w^{T}5x^i+3b)}}$

Loss function:

$J(w,b)=\frac{1}{2m}{\displaystyle \underset{i=1}{\overset{m}{}}}\{(y^i-(hat(y^i)){)}^2\}$

Group $2$:

Activation function:

$a_{w,b}(x^i)=\frac{1}{1+e^{-(w^T{x}^i+b)}}$

Loss function:

$J(w,b)=\frac{1}{m}{\displaystyle \underset{i=1}{\overset{m}{}}}\{(y^i-$hat$(y{)}^i{)}^2\}$

Testing these two implemented Perceptrons on the testing dataset $($test$\text{'}$catvnoncat$.$h$5),$ respectively, with the same hyper$-$parameters: learning rate: $0\mathrm{.}01,$ iterations: $2000.$

Calculate the training accuracies and compare their differences

Calculate the testing accuracies and compare their differences

Plot the learning curves for these two Perceptron, respectively and compare and analyze their differences.

Requirements:

Submitting the source code $($Jupyter notebook files from Google Colab$)$ and the data you use for the project, where the codes have no bugs.

Write and submit the project report with the following parts:

Project description

Model description.

Result analysis