Class 10 AI Chapter - Modelling in AI Topic - Deep Learning

 

Class 10 Artificial Intelligent Code 417 Solutions

Session 2025-26

Artificial Intelligence code 417 syllabus pdf class 10 solutions. Class 10 AI Book. Class 10 AI Notes. This article provides complete solution for class 10 AI (Artificial Intelligence)  Code - 417 students according to new syllabus 2025 – 26. In this article first of all go through the AI Code - 417 syllabus and then follow chapter wise notes.

--------------------------------------------------

Chapter - Modelling in AI
Other Topics

• AI ML DL
• AI Modelling
• Machine Learning
• Deep Learning  
  

-------------------------------------------------

Deep Learning

Deep Learning is a specialized branch of machine learning that uses artificial neural networks (ANNs) with multiple layers to simulate how humans learn. It allows computers to learn directly from raw data such as images, sound, or text, without manually programming rules.

• Works best with large datasets and high computational power.
• Automatically extracts features and patterns from data.
• Used in voice assistants, self-driving cars, medical imaging, and more.
  
  

  

  AI Model

  
  

1. Artificial Neural Network (ANN)

What is ANN?

An Artificial Neural Network is a computing system inspired by the human brain. It consists of layers of nodes (also called neurons) connected to each other, capable of learning from data.

Structure of ANN:

1. Input Layer:

   • Takes raw data (e.g., pixel values, numerical features).
     
     

2. Hidden Layers:

   • Multiple layers that process and transform input using weights and biases.
   • Each neuron applies an activation function to decide output.
     
     

3. Output Layer:

   • Produces the final prediction (e.g., classification or regression).
     
     

How ANN Works (Simplified):

1. Each neuron receives input, multiplies it by a weight, adds a bias.

2. The result goes through an activation function like ReLU or Sigmoid.

3. ANN adjusts weights through backpropagation using gradient descent.

Example Use Cases:

• Fraud detection in banking
• Predicting stock prices
• Handwriting recognition
  
  

2. Convolutional Neural Network (CNN)

What is CNN?

A Convolutional Neural Network is a type of deep neural network specifically designed for image and spatial data. It excels at capturing patterns like edges, textures, and shapes.

Structure of CNN:

1. Input Layer:
   

   • Accepts image data (e.g., 28x28 pixels)
     
     

2. Convolutional Layers:
   

   • Applies filters (kernels) to extract features from images
   • Each filter detects patterns like edges or corners
     
     

3. ReLU Layer (Activation):
   

   • Applies the ReLU function to introduce non-linearity
     
     

4. Pooling Layer:
   

   • Reduces image dimensions (e.g., Max Pooling)
   • Makes model faster and reduces over fitting
     
     

5. Fully Connected Layers (Dense):
   

   • Final decision-making layers for classification

6. Output Layer:
   

   • Gives predicted result (e.g., Cat or Dog)
     

Example Use Cases:

• Face recognition
• Medical image diagnosis
• Self-driving car vision
• Object detection (e.g., identifying people in photos)
  
  

ANN vs CNN:

Feature	ANN	CNN
Data Type	Tabular, simple data	Images, spatial data
Layers	Dense (fully connected) layers	Convolution + Pooling + Dense
Feature Learning	Manual or basic	Automatic feature extraction
Accuracy in Images	Lower	High

---

How Does AI Make a Decision

Step-by-Step Neural Decision Process

1. Inputs (x)

These are the features or values from the environment or dataset.

Example: For an AI predicting house price:

• x₁ = size of house
• x₂ = number of bedrooms
• x₃ = age of house

2. Weights (w)

Each input is multiplied by a weight, which tells the AI how important that input is.

Example:

• w₁ = 0.6
• w₂ = 0.2
• w₃ = 0.1

3. Weighted Sum

All weighted inputs are added together along with a bias (b).

$$z = (x₁ \cdot w₁) + (x₂ \cdot w₂) + (x₃ \cdot w₃) + b$$

The bias allows the AI to shift the result up or down, giving flexibility.

4. Activation Function

The result z is passed through an activation function to introduce non-linearity — this helps the AI learn complex patterns.

Common activation functions:

• Sigmoid → Outputs between 0 and 1
• ReLU (Rectified Linear Unit) → Outputs 0 if input < 0, else outputs the input
• Tanh → Outputs between -1 and 1

$$$$Output=Activation(z)

5. Final Output

After activation, we get the neuron's final output — a decision like:

• Classifying an image
• Predicting a value
• Approving a loan, etc.
  
  

Example:

Goal: Predict if a person will buy a product (Yes = 1, No = 0)

We'll use 2 inputs:

• x₁ = Number of ads seen by the person
• x₂ = Time spent on website (in minutes)

Given:

• Inputs: x₁ = 3, x₂ = 5
• Weights: w₁ = 0.4, w₂ = 0.6
• Bias: b = 1
• Activation Function: Sigmoid
  

  

  Perception Model

Step-by-Step

1. Weighted Sum

$$$$z=(x1⋅w1)+(x2⋅w2)+b=(3⋅0.4)+(5⋅0.6)+1=1.2+3.0+1=5.2

2. Activation Function (Sigmoid)

Sigmoid formula:

$$$$Sigmoid(z)=11+e−z=11+e−5.2≈0.9945

3. Final Output

$$$$Output=0.9945≈1\text{Output} = 0.9945 \approx 1

This means the AI predicts the person will buy the product.

Great question, Alvin! 😊

In the sigmoid function formula:

$$$$Sigmoid(z)=11+e−z\text{Sigmoid}(z) = \frac{1}{1 + e^{-z}}

What is e?

• e is a mathematical constant called Euler's number.
• Its approximate value is:          $$e\approx 2.71828$$$$$$It’s used in exponential functions, and it's fundamental in calculus, probability, and neural networks.$$$$
  

---