Implement Neural Network in Python from Scratch ! In this video, we will implement MultClass Classification with Softmax by making a Neural Network in Python from Scratch. We will not use any build in models, but we will understand the Mathematics and Code behind the Neural Network in Python.

IMPORTANT NOTE

I made a small mistake while typing the code for derivative_tanh(x) function.

The correct code will be :
def derivative_tanh(x):
return 1 - np.power(np.tanh(x), 2)

And not:
def derivative_tanh(x):
return 1 - np.power(x, 2)

I have explained it in this video at 19:05 : https://www.youtube.com/watch?v=URJ9pP1aURo&list=PLuhqtP7jdD8CftMk831qdE8BlIteSaNzD&index=5



Download the ASSIGNMENT and Implementation Code : https://github.com/Coding-Lane/Neural-Network---MultiClass-Classifcation-with-Softmax



Complete Neural Network Playlist : https://www.youtube.com/watch?v=mlk0rddP3L4&list=PLuhqtP7jdD8CftMk831qdE8BlIteSaNzD



LEARNINGS : Why did Cost increased at the beginning of training the model?

Answer : It is because our weights W were very large. We multiplied 0.01 with them, but still it was not enough.

Solution : Multiply 0.001 with weights W, instead of 0.01. You will that problem will be solved and you will have a steady decreasing cost function.

There are many weight initialization techniques in Deep Learning to solve such problems and more. We will discuss them in the upcoming videos. We will see why is it necessary to initialize weights randomly and not with zeros. And also why weights need to be small !



How Neural Network works ? : https://www.youtube.com/watch?v=mlk0rddP3L4&list=PLuhqtP7jdD8CftMk831qdE8BlIteSaNzD&index=1

Forward and Backpropagation in Neural Network : https://www.youtube.com/watch?v=Tb23YtZ92AE&list=PLuhqtP7jdD8CftMk831qdE8BlIteSaNzD&index=2

Activation Functions in Neural Network : https://www.youtube.com/watch?v=Y9qdKsOHRjA&list=PLuhqtP7jdD8CftMk831qdE8BlIteSaNzD&index=3

Types of Cost Functions in Neural Network : https://www.youtube.com/watch?v=NJpABYQB9PI&list=PLuhqtP7jdD8CftMk831qdE8BlIteSaNzD&index=4

Derivation of Equations in Backpropagation : https://www.youtube.com/watch?v=URJ9pP1aURo&list=PLuhqtP7jdD8CftMk831qdE8BlIteSaNzD&index=5

Backpropagation in MultiClass Classification with Softmax : https://www.youtube.com/watch?v=f-nW8cSa_Ec&list=PLuhqtP7jdD8CftMk831qdE8BlIteSaNzD&index=6



0:00 Intro and Agenda
1:41 Importing Dataset
5:18 Neural Network summary
7:10 Activation Function
8:55 Initializing Parameters
10:20 Forward Propagation Implementation
11:48 Cost Function Implementation
12:58 Backpropagation Implementation
15:40 Update Parameters Implementation
16:28 Complete Model Implementation
22:01 Checking Accuracy
25:02 Making Predictions
25:32 Ending



Learn what is Supervised Learning : https://www.youtube.com/watch?v=Bo5dJT1QlHc&list=PLuhqtP7jdD8BpW2kOdIbjLI3HpuqeoMb-&index=2

Complete Neural Network playlist : https://www.youtube.com/watch?v=U1omz0B9FTw&list=PLuhqtP7jdD8Chy7QIo5U0zzKP8-emLdny

Complete Logistic Regression Playlist : https://www.youtube.com/watch?v=xJjr_LPfBCQ&list=PLuhqtP7jdD8BpW2kOdIbjLI3HpuqeoMb-



This is Your Lane to Machine Learning



Subscribe to my channel, because I upload a new Machine Learning video every week : https://www.youtube.com/channel/UCJFAF6IsaMkzHBDdfriY-yQ?sub_confirmation=1