Unit 2: Advanced Concepts of Modeling in AI
1. Define AI, ML, DL
AI (Artificial Intelligence): Enables machines to mimic human intelligence and perform tasks like decision-making.
ML (Machine Learning): A subset of AI that enables systems to learn and improve from experience.
DL (Deep Learning): A further subset of ML using artificial neural networks to process vast data and perform complex tasks.
2. What is Data?
Data is information in any form, used for reference or analysis. For example, a table listing fruit names, colors, and sizes is data.
3. What do you mean by Features?
Features are individual measurable properties or characteristics of data. In a fruit dataset, features may include color, size, and weight.
4. What are Labels?
Labels are the outputs or target values in a dataset. They help in predicting outcomes. For example, fruit names in a dataset used to classify by color.
5. What do you mean by Training Data and Testing Data?
Training Data is used to teach a model (contains features and labels).
Testing Data is used to evaluate the model’s accuracy using new, unseen data.
6. Differences between Rule-based and Learning-based AI models
| Rule-Based | Learning-Based |
| Uses fixed rules defined by the developer | Learns patterns from data |
| cannot learn from new data | adapts over time |
| static behavior, cannot adapt to changes | dynamic behavior |
| Ex : chatbot with pre-defined rules | Ex: Spam filters that learn from emails |
7. What are the different types of ML Explain?
Ans : Different types of ML are: Supervised, Unsupervised and Reinforcement Learning
🔹 Supervised Learning:
In supervised learning, the model is trained on labeled data (data with input and correct output). It’s like a student learning from solved examples and then answering similar questions.
Example: Spam detection in emails – the system learns from emails labeled as ‘spam’ or ‘not spam’.
🔹 Unsupervised Learning:
Here, the model works with unlabeled data and finds hidden patterns or groupings. It’s like exploring and organizing information without knowing the categories beforehand.
Example: Grouping customers based on purchase behavior without any predefined labels.
🔹 Reinforcement Learning:
The model learns by interacting with the environment, receiving feedback as rewards or penalties. It’s like learning by trial and error.
Example: A robot learning to park a car gets a reward when it parks correctly and a penalty for crashing.
Q1. What is clustering and how is it different from classification?
Clustering is a type of learning where the machine groups similar data together without using any labels. It finds hidden patterns and divides the data into clusters based on similarity.
Classification is when the machine is trained using labeled data and then it puts new data into those known labels or categories. In short, classification uses known labels, while clustering does not.
Q2. Explain neural networks. Also give functions of three layers of neural networks.
A neural network is a model that works like the human brain. It helps machines learn and take decisions, especially with large and complex data.
Functions of 3 layers:
1. Input Layer: This layer takes the input data (like images or numbers).
2. Hidden Layer: It processes the input using some rules and calculations.
3. Output Layer: It gives the final answer or prediction based on processed data.
Q3. Differentiate between classification and regression model.
Classification model is used to predict categories like Yes/No or spam/not spam. It gives a fixed set of outputs.
Regression model is used to predict continuous values like age, price, or temperature. It gives a number as the answer.
Q4. Name and explain two types of Unsupervised Learning Models with example
Unsupervised learning means the machine is given data without labels, and it learns to find patterns or relationships on its own.
Clustering
Clustering is a type of unsupervised learning where a machine groups similar data items together.
It does this without using any labels — the machine finds patterns in the data on its own.
Example: Grouping customers based on shopping habits or grouping animals based on size and color.
Association
Association is a type of unsupervised learning that finds relationships between items in a dataset.
It is used to discover which items are often bought or used together.
Example: If people who buy bread also buy butter, the machine will learn that “bread → butter” is a strong association.
Q5. Name and explain sub-categories of Deep Learning
ANN (Artificial Neural Network)
An Artificial Neural Network (ANN) is a type of deep learning model that works like the human brain.
It has layers of interconnected nodes (like brain cells) that help process information and make decisions.
ANNs are used for tasks like speech recognition, weather prediction, and handwriting recognition.
CNN (Convolutional Neural Network)
A Convolutional Neural Network (CNN) is a special kind of neural network mainly used for image-related tasks.
CNNs can automatically detect features like shapes, edges, and patterns in pictures.
They are commonly used in facial recognition, medical image analysis, and self-driving cars.
Q6. Real-world applications of Neural Networks
1. Face recognition in phones
2. Chatbots for customer support
3. Predicting vegetable prices
4. Voice assistants like Alexa
Q7. What is Perceptron? Explain
A Perceptron is the most basic building block of a neural network. It is used to make decisions by taking inputs and checking if they are enough to give an output.
How it works:
– Takes inputs and gives each a weight (importance).
– Adds all and compares the result to a threshold.
– If total > threshold → output is 1 (Yes). If not → output is 0 (No).
Example: Deciding to go out based on weather and items like umbrella or jacket.