Class 10 AI Chapter - Modelling in AI Topic - Question/Answer - Arvindzeclass

Class 10 Artificial Intelligent Code 417 Solutions

Session 2025-26

Artificial Intelligence code 417 syllabus pdf class 10 solutions. Class 10 AI Book. Part - A, Part - B, and Python. This article provides complete solution for class 10 AI (Artificial Intelligence) Code - 417 solution 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. Class 10 AI MCQs Quiz with answers.

Part A - Employability Skills Question/Answers

Chapter 1 -Communication Skills - II

Chapter 2 -Self - Management Skills - II

Chapter 3 - ICT Skills - II

Chapter 4 - Entrepreneurial Skills - II

Chapter 5 - Green Skills - II

Part B - Subject Specific Skills Question/Answers

Chapter 1 -AI Project Cycle & Ethics

Chapter 2 - Modelling in AI

Question Answer - 1
Question Answer - 2
Question Answer - 3
Question Answer - 4

Chapter 3 - Evaluating Models

Chapter 4 - Statistical Data

Chapter 5 - Computer Vision

Chapter 6 - Natural Language Processing

Chapter 7 - Advance Python

#What is Training and Testing Dataset?

When we build a machine learning (ML) model, we usually divide our data into two main parts:

1. Training Dataset

Definition: The portion of the data used to train the model.
Purpose: The model learns patterns, relationships, and features from this data.
Example: Suppose you’re teaching an ML model to recognize fruits. The training dataset will have many labeled examples like:
🍎 Apple → "Apple"
🍌 Banana → "Banana"
🍊 Orange → "Orange"

The model uses these examples to learn the mapping between inputs (fruit images) and outputs (labels).

Dataset Split

2. Testing Dataset

Definition: The portion of the data used to evaluate how well the model has learned.
Purpose: It checks whether the model can generalize to new, unseen data.
Example: After training, you test the model on different fruit images (not used during training) to see if it predicts correctly:
🍎 Unknown apple → Model says "Apple" ✅
🍌 Unknown banana → Model says "Banana" ✅
🍐 Unknown pear → Model says "Apple" ❌

Key Points

Usually, we split data into 70–80% training and 20–30% testing.
Sometimes, we also use a Validation dataset (to tune parameters before final testing).
Goal: Ensure the model doesn’t just memorize the training data (overfitting) but can perform well on new data.

👉 In short:

Training dataset = used to teach the model.
Testing dataset = used to check the model’s performance on unseen data.

#What is Rule-based and Learning-based approach?

1. Rule-Based Approach

Definition: In this approach, a system works based on predefined rules written by humans.
These rules are usually in the form of if-then statements.
The system does not learn from data — it only follows the rules given.

Example:
👉 Spam email detection (Rule-based)

If the email contains words like “lottery” or “free money” → mark as spam.
If the sender is in contacts → mark as not spam.

📌 Advantages:

Easy to understand.
Works well for simple, structured problems.

📌 Disadvantages:

Cannot handle complex or new situations.
Requires constant updating by humans.
Not scalable for large datasets.

2. Learning-Based Approach

Definition: In this approach, the system learns patterns from data instead of relying only on predefined rules.
Uses machine learning (ML) or deep learning (DL) techniques.
The system improves its performance as it sees more data.

Example:
👉 Spam email detection (Learning-based)

Collect many examples of spam and non-spam emails.
Train a machine learning model (e.g., Naive Bayes, Neural Network).
The model automatically learns patterns like suspicious words, frequency, or sender behavior.
It can detect new spam emails it hasn’t seen before.

📌 Advantages:

Can handle large and complex data.
Improves automatically as more data is added.
More accurate than rule-based in real-world applications.

📌 Disadvantages:

Requires large amounts of data.
Needs computational power.
Sometimes acts like a “black box” (hard to understand how it makes decisions).

Key Difference

Feature	Rule-Based Approach	Learning-Based Approach
How it works	Follows predefined rules (if-then)	Learns patterns from data
Flexibility	Low (needs manual updates)	High (improves with new data)
Scalability	Not scalable for large problems	Very scalable with big data
Example	Spam filter with keywords	Spam filter using ML model

👉 In simple terms:

Rule-based = like giving strict instructions.
Learning-based = like teaching the system with examples.

#What is Machine Learning?
Explain Supervised, Unsupervised, and Reinforcement Learning.

Machine Learning (ML)

Definition: Machine Learning is a branch of Artificial Intelligence (AI) where computers learn patterns from data and make predictions or decisions without being explicitly programmed with rules.
Instead of giving instructions step by step, we give the system data and examples, and it learns by itself.

👉 Example:

You show a machine thousands of pictures of cats and dogs.
It learns the patterns (like ears, tails, whiskers) and can later predict whether a new picture is a cat or dog.
AI Model

Types of Machine Learning

1. Supervised Learning

Definition: The model is trained on a dataset that has both inputs and outputs (labels).
The system learns a mapping from input → output.
After training, it can predict the output for unseen inputs.

Example:

Input: Features of a house (size, location, rooms).
Output (label): House price.
Model learns from past data and predicts new house prices.

📌 Algorithms: Linear Regression, Decision Trees, Neural Networks.

2. Unsupervised Learning

Definition: The dataset has only inputs, no labels.
The system tries to find hidden patterns or groupings in the data.

Example:

Input: Customer purchase history.
No output labels.
Model groups customers into clusters (e.g., frequent buyers, occasional buyers, one-time buyers).

📌 Algorithms: K-Means Clustering, Hierarchical Clustering, PCA (Dimensionality Reduction).

3. Reinforcement Learning (RL)

Definition: The model learns by interacting with an environment.
It receives rewards or penalties based on actions.
The goal is to maximize long-term rewards by learning the best strategy (policy).

Example:

A robot learning to walk.
Each time it falls → negative reward (penalty).
Each step forward → positive reward.
Over time, it learns the best way to walk.

📌 Algorithms: Q-Learning, Deep Q-Networks (DQN), Policy Gradient.

Comparison Table

Type	Data Used	Goal	Example
Supervised	Input + Output (labeled data)	Predict outcomes	Predict house prices
Unsupervised	Input only (no labels)	Find patterns/clusters	Customer segmentation
Reinforcement	Interaction + Feedback	Learn best actions (policy)	Training robots / Game AI

👉 In simple words:

Supervised → Teacher gives questions and answers to learn.
Unsupervised → No teacher, student finds patterns by themselves.
Reinforcement → Learn by trial and error with rewards and penalties.

#What is Deep Learning?
Explain Artificial Neural Network and Convolutional Neural Network.

🌟Deep Learning

Deep Learning (DL) is a part of Machine Learning.
It uses a special kind of model called a Neural Network, which is designed to work like the human brain.
These networks can automatically learn very complex patterns from a huge amount of data.

👉 Example:

Machine Learning may identify a fruit if you give it features like color, size, weight.
Deep Learning automatically finds features (like edges, shapes, textures) from raw images of the fruit—no manual feature design needed.

That’s why deep learning is powerful for:

Face recognition
Self-driving cars
Voice assistants (Alexa, Siri)
Medical image analysis

🧠 Artificial Neural Network (ANN)

1. What is ANN?

ANN is the basic structure of deep learning models.
Inspired by the human brain, where neurons (nerve cells) connect to pass signals.
ANN consists of layers of artificial neurons that process data step by step.

2. Structure of ANN

Input Layer: Takes raw data (e.g., pixels of an image, marks of a student).
Hidden Layers: Process the data through many mathematical operations and transformations.
Output Layer: Gives the final prediction (e.g., cat/dog, pass/fail, price).

👉 Example:
If you give a student’s marks in subjects as input:

Input layer → marks (Math, Science, English).
Hidden layers → detect patterns (good in science, weak in math).
Output layer → predicts if student will pass or fail.

3. How ANN learns

Data goes forward through layers (Forward Propagation).
Computer checks the error between prediction and actual answer.
It sends correction signals backward (Back propagation) to improve the network.
With many cycles, ANN learns to give accurate results.

🖼️ Convolutional Neural Network (CNN)

1. What is CNN?

CNN is a special type of Neural Network, mainly used for images and visual data.
It is very good at detecting patterns like edges, shapes, textures automatically.

2. Structure of CNN

Convolutional Layer: Applies filters (small windows) that scan the image to detect edges, corners, colors, etc.
Pooling Layer: Reduces size of data but keeps important features (like shrinking the picture but keeping the shape).
Fully Connected Layer: Works like an ANN’s output layer, combining all features to make the final decision.

3. Example

Imagine recognizing if a picture has a cat or dog:

Convolution layer: detects ears, eyes, nose, whiskers.
Pooling layer: reduces picture size, keeps main features.
Fully connected layer: decides “Cat” or “Dog.”

📊 ANN vs CNN (Comparison)

Feature	ANN	CNN
Data type	Works with numbers, tables, simple features	Best for images, videos, and visual data
Feature extraction	Manual (we must choose features)	Automatic (learns features itself)
Layers	Input → Hidden → Output	Convolution → Pooling → Fully Connected
Example use	Predict exam results, house prices	Face recognition, object detection

👉 Summary in Simple Words

Deep Learning = advanced machine learning using neural networks.
ANN = like the brain, processes data step by step through layers.
CNN = a type of ANN, very strong in image/video recognition because it automatically finds patterns.