Introduction to Artificial Intelligence, ML, DL and NLP

 

Introduction to Artificial Intelligence, Machine Learning, Deep Learning and Natural Language Processing

---

1. Ice Breaker 

• How many of you use ChatGPT?

• How does Instagram recommend reels?

• How does Google Maps predict traffic?

• How does Face Unlock work?

• How does Alexa understand your voice?

All these technologies are powered by Artificial Intelligence, Machine Learning, Deep Learning and Natural Language Processing.

---

2. Evolution of Artificial Intelligence

Draw on Board

Artificial Intelligence (AI)
│
├── Machine Learning (ML)
│   ├── Supervised Learning
│   ├── Unsupervised Learning
│   └── Reinforcement Learning
│
├── Deep Learning (DL)
│   ├── Artificial Neural Networks (ANN)
│   ├── Convolutional Neural Networks (CNN)
│   ├── Recurrent Neural Networks (RNN)
│   ├── Long Short-Term Memory Networks (LSTM)
│   ├── Gated Recurrent Units (GRU)
│   └── Transformers
│
├── Natural Language Processing (NLP)
│
├── Computer Vision
│
├── Robotics
│
├── Expert Systems
│
└── Knowledge Representation

Explain:

• Artificial Intelligence is the parent field.

• Machine Learning is a subset of Artificial Intelligence.

• Deep Learning is a subset of Machine Learning.

• Natural Language Processing is a branch of Artificial Intelligence focused on human language.

---

3. What is Artificial Intelligence (AI)?

Definition

Artificial Intelligence is the capability of a machine to mimic human intelligence such as:

• Learning

• Reasoning

• Decision Making

• Problem Solving

• Understanding Language

• Recognizing Images

Examples

• ChatGPT

• Siri

• Alexa

• Google Assistant

• Self-driving cars

• Face recognition systems

• Fraud detection systems

---

4. What is Machine Learning (ML)?

Definition

Machine Learning is a technique where computers learn patterns from data and improve their performance without being explicitly programmed.

Traditional Programming

Data + Program → Output

Machine Learning

Data + Output → Model

New Data + Model → Prediction

Examples

• House Price Prediction

• Student Placement Prediction

• Loan Approval Prediction

• Employee Attrition Prediction

• Customer Churn Prediction

• Sales Forecasting

• Stock Market Trend Prediction

• Credit Card Fraud Detection

---

5. What is Deep Learning (DL)?

Definition

Deep Learning is a specialized branch of Machine Learning that uses Artificial Neural Networks with multiple hidden layers to learn complex patterns from large datasets.

Why Deep Learning is Growing?

Discuss:

1. Availability of Big Data

2. High-performance Graphics Processing Units (GPUs)

3. Cloud Computing

4. Better Algorithms

5. Increased Storage Capacity

6. Internet and Social Media Data

7. Open-source frameworks

Examples:

• TensorFlow

• PyTorch

• Keras

---

6. What is Natural Language Processing (NLP)?

Definition

Natural Language Processing is a branch of Artificial Intelligence that enables computers to understand, process, analyze and generate human language.

Languages Examples

• English

• Marathi

• Hindi

• French

• Japanese

Why NLP is Required?

Computers understand numbers, not human language.

NLP converts:

Human Language
       ↓
Machine Understandable Form
       ↓
Prediction / Analysis / Response

---

7. Applications of Natural Language Processing (NLP)

Text Applications

• Email Classification

• Spam Detection

• Document Classification

• Resume Screening

• Text Summarization

• News Categorization

Language Applications

• Language Translation

• Grammar Correction

• Spell Checking

• Auto Completion

Sentiment Analysis

• Customer Reviews

• Student Feedback Analysis

• Product Reviews

• Social Media Analysis

Conversational AI

• ChatGPT

• Gemini

• Claude

• Customer Service Chatbots

• Virtual Assistants

Healthcare

• Clinical Report Analysis

• Medical Record Summarization

Legal Domain

• Contract Analysis

• Legal Document Classification

---

8. Difference Between Machine Learning and Deep Learning

Parameter	Machine Learning (ML)	Deep Learning (DL)
Data Requirement	Small to Medium	Very Large
Feature Engineering	Manual	Automatic
Training Time	Faster	Slower
Hardware Requirement	Central Processing Unit (CPU) sufficient	Graphics Processing Unit (GPU) preferred
Complexity	Lower	Higher
Interpretability	Easier	Difficult
Accuracy	Good	Very High
Best For	Structured Data	Images, Audio, Video, Text

---

9. Applications Where Only Machine Learning is Commonly Used

Education

• Student Placement Prediction

• Student Performance Prediction

• Dropout Prediction

Banking

• Loan Approval

• Credit Risk Analysis

• Customer Segmentation

Marketing

• Customer Churn Prediction

• Sales Forecasting

• Demand Forecasting

Human Resources

• Employee Attrition Prediction

• Recruitment Analytics

Healthcare

• Diabetes Prediction

• Heart Disease Prediction

Insurance

• Claim Prediction

• Risk Assessment

---

10. Applications Where Deep Learning is Preferred

Computer Vision

• Face Recognition

• Face Detection

• Object Detection

• Image Classification

• Medical Imaging

Autonomous Vehicles

• Self-driving Cars

• Lane Detection

• Traffic Sign Recognition

Speech Processing

• Speech Recognition

• Voice Assistants

• Speaker Identification

Video Analytics

• CCTV Surveillance

• Activity Recognition

• Crowd Monitoring

Generative Artificial Intelligence

• ChatGPT

• Gemini

• Claude

• Image Generation

• Video Generation

Healthcare

• Cancer Detection from MRI Scans

• Tumor Detection

• X-Ray Analysis

Cyber Security

• Malware Detection

• Intrusion Detection

---

11. Real-Life Examples 

Application	Technology
ChatGPT	DL + NLP
Face Unlock	DL
Google Translate	DL + NLP
Student Placement Prediction	ML
Netflix Recommendation	ML
YouTube Recommendation	ML + DL
Speech Recognition	DL
Spam Detection	ML + NLP
Sentiment Analysis	NLP
Fraud Detection	ML
Medical Image Analysis	DL
Self-Driving Car	DL

---

Deep Learning:  Real-World Applications

---

 Why Deep Learning Was Needed? 

Problem 1: Human Face Recognition

Ask:

Can you identify your friend in a group photo?

Students: Yes

Ask:

How do you do it?

Possible answers:

• Eyes

• Nose

• Hair

• Face shape

Explain:

Humans automatically learn features.

Traditional Machine Learning cannot automatically identify these features.

A Data Scientist must manually provide features:

Face Length
Eye Distance
Nose Width
Skin Tone
Hair Length

This is called:

Feature Engineering

Problem:

• Time consuming

• Human dependent

• Not scalable

Deep Learning automatically learns features.

---

Evolution of Artificial Intelligence 

Draw:

Artificial Intelligence
       │
       ▼
Machine Learning
       │
       ▼
Deep Learning

Historical Timeline

Year	Development
1950	Turing Test
1956	Birth of AI
1958	Perceptron
1986	Backpropagation
1997	IBM Deep Blue
2012	AlexNet Revolution
2017	Transformer Architecture
2022	ChatGPT
Today	Generative AI

Explain:

Deep Learning existed earlier but became successful due to:

• Big Data

• Graphics Processing Units (GPU)

• Cloud Computing

• Better Algorithms

---

What Exactly is Deep Learning? 

Definition:

Deep Learning is a subset of Machine Learning that uses Artificial Neural Networks containing multiple hidden layers to learn complex patterns from data.

Draw:

Input Layer
      │
Hidden Layer 1
      │
Hidden Layer 2
      │
Hidden Layer 3
      │
Output Layer

Explain:

The word "Deep" means:

More Hidden Layers

Not bigger data.

---

How Deep Learning Learns 

Take an example:

Cat vs Dog Classification

Input:

Image

Layer 1 learns:

Edges
Lines
Corners

Layer 2 learns:

Eyes
Nose
Ears

Layer 3 learns:

Face Structure

Output:

Dog
or
Cat

Explain:

Deep Learning learns from simple to complex.

Exactly like a child learns.

---

Real-World Applications 

Application 1: Face Unlock

Problem:

Phone should unlock only for owner.

Solution:

Deep Learning identifies facial patterns.

Used by:

• Apple Face ID

• Samsung Face Recognition

---

Application 2: Self Driving Cars

Problem:

Vehicle must understand surroundings.

Tasks:

• Detect pedestrians

• Detect signals

• Detect vehicles

• Detect lanes

Deep Learning Models:

• Convolutional Neural Networks (CNN)

---

Application 3: Medical Diagnosis

Problem:

Detect cancer early.

Input:

MRI Scan

Output:

Cancer or No Cancer

Advantage:

Often comparable to expert radiologists.

---

Application 4: ChatGPT

Problem:

Understand and generate human language.

Input:

What is Deep Learning?

Output:

Meaningful response

Uses:

• Transformers

• Large Language Models (LLMs)

---

Application 5: Google Translate

Input:

Hello

Output:

नमस्कार

Deep Learning understands context.

---

Application 6: YouTube Recommendations

Problem:

Suggest relevant videos.

Deep Learning analyzes:

• Watch History

• Search History

• Watch Duration

• Likes

---

Application 7: Fraud Detection

Problem:

Millions of transactions every day.

Deep Learning identifies:

• Unusual spending

• Suspicious patterns

• Potential fraud

---

Application 8: Agriculture

Problem:

Detect crop disease.

Input:

Leaf Image

Output:

Healthy or Diseased

---

Application 9: Smart Cities

Applications:

• Traffic Monitoring

• Vehicle Counting

• Number Plate Recognition

• Crowd Detection

---

Application 10: Cyber Security

Applications:

• Intrusion Detection

• Malware Detection

• Phishing Detection

• Threat Prediction

---

Deep Learning vs Human Brain

Human Brain
     ↓
Biological Neurons

Deep Learning
     ↓
Artificial Neurons

Explain:

Human Brain	Deep Learning
Neurons	Artificial Neurons
Synapses	Weights
Learning	Training
Memory	Model Parameters

Homework / Discussion Task

Ask students to find:

One Application Each Of:

1. Deep Learning in Healthcare

2. Deep Learning in Education

3. Deep Learning in Banking

4. Deep Learning in Agriculture

5. Deep Learning in Cyber Security

and explain:

• Problem

• Input Data

• Deep Learning Model Used

• Output