Artificial Intelligence

x1, x2, x3 are input for neural network taken from the first row in the dataset

Ytarget is from the first row in the dataset

a = 1 / (1 + e-z)

- is a igmoid function

- to squeeze numbers into range between 0-1

edit: please fill the blanks at the last image (Feedforward, Backpropagation, and Calculate the final weights)

As a data scientist working for an IT security company, you are asked to create a program using machine learning model that could classify if a web page is potentially a scam/fraud website or not. You decided to use neural network to solve the problem. The neural network receives three input variables and provide one output ( scam =1, safe =0 ). Input Variable: - fake subdomain: if a website contains a sub-domain then value =1, otherwise =0 - asking payment: number of times the web page mention words such as payment/subscribe/fee/etc. - contain typo: number of words with typos on a web page. with the following information on initial weights: W1=0.2W3=0.8W2=0.5W4=0.4 learning rate =1 Tasks: 1. Perform Feedforward 2. Perform Backpropagation 3. Calcualte the final Weights There are two processes to calculate final weights: 1. Feedforward 2. Backpropagation 1. Feedforward Next, perform activation function z=sum(xiweight)H3=(x1W1)+(x2w2)+(x3w3)H3=tosqueezenumbersintorange01a=1/(1+ez)H3=1/(1+e)H3= z=sum(xiweight)O4=(H3W4)O4=()O4=a=1/(1+ez)O4=1/(1+e)O4=(NetworkOutput(y))Error=ytargety=