Name: PptxGenJS Presentation
Uploaded: 2026-09-06
Duration: 9 min 18 s
Description: CHAPTER 1. Introduction to Machine Learning. AI · ML · Deep Learning · Workflows · Applications.

CHAPTER 1. Introduction to Machine Learning. AI · ML · Deep Learning · Workflows · Applications. 

Scene 1

AI · ML · Deep Learning · Workflows · Applications

Today's Lecture — What We Cover. 1. Why ML?. Traditional vs ML programming, Big Data era. 

Scene 2

Traditional vs ML programming, Big Data era

SECTION 1. Why Machine Learning?. From rule-based programming to data-driven intelligence. 

Scene 3

From rule-based programming to data-driven intelligence

SECTION 1 · WHY ML?. Traditional Programming vs Machine Learning. 

Scene 4

Traditional Programming vs Machine Learning

You write the rules manually. Works for simple tasks — breaks for complex ones.

The computer learns the rules from data. Scales to complex, real-world problems.

SECTION 1 · WHY ML?. Why ML Matters in the Big Data Era. 

Scene 5

Billions of records are generated every second — too much for manual analysis.

Patterns in big data are far too intricate for humans to detect by writing rules.

ML enables real-time predictions and decisions at machine speed, at scale.

Organizations that leverage ML outperform those relying only on traditional methods.

SECTION 1 · WHY ML?. ML Is Already Part of Your Daily Life. 

Scene 6

Ask yourself: How does YouTube know what video you might like?

SECTION 2. AI, Machine Learning & Deep Learning. Understanding the hierarchy of intelligent systems. 

Scene 7

Understanding the hierarchy of intelligent systems

SECTION 2 · AI / ML / DL. The Hierarchy: AI ⊃ ML ⊃ Deep Learning. 

Scene 8

The Hierarchy: AI ⊃ ML ⊃ Deep Learning

The broadest concept: systems that perform tasks associated with human intelligence — reasoning, planning, language understanding, perception, decision-making.

A major approach within AI where systems learn patterns from data, rather than relying on explicitly programmed rules. Examples: spam detection, house price prediction.

A subset of ML using multi-layer neural networks. Powers image recognition, speech recognition, NLP, and generative AI.

💡 CLASSROOM DISCUSSION. "Is every AI system a Machine Learning system?". 

Scene 9

"Is every AI system a Machine Learning system?"

Uses pre-programmed rules. No learning from data.

Learns patterns from data. Improves with experience.

Not all AI is ML — but ML is a powerful subset of AI!

SECTION 3. What Is Machine Learning?. Definitions, terminology, and the learning concept. 

Scene 10

Definitions, terminology, and the learning concept

SECTION 3 · WHAT IS ML?. Defining Machine Learning. 

Scene 11

"Machine Learning is a method of teaching computers to learn patterns from data and use those patterns to make predictions or decisions."

A machine does not learn exactly like a human — it discovers mathematical relationships in data.

SECTION 3 · WHAT IS ML?. Key Machine Learning Terminology. 

Scene 12

The entire collection of data used for training and testing.

One single row or data point in the dataset.

An input variable used to make predictions (e.g., size, age).

The output we want to predict (e.g., price, category).

The mathematical procedure used to learn from data.

The learned output of the algorithm — used to make predictions.

The process of feeding data to the algorithm so it learns.

The model's output when given new, unseen data.

Measuring how well the model performs on test data.

SECTION 3 · WHAT IS ML?. Example: House Price Prediction. 

Scene 13

Feature Role Example Value 📐  Size Input Feature 85 m² 🛏️  Bedrooms Input Feature 3 rooms 📍  Location Input Feature Seoul 📅  Age Input Feature 8 years 💰  Price TARGET (Label) ₩650 million

Size + Bedrooms + Location + Age = Features

SECTION 4. How Does Machine Learning Work?. The learning pipeline from raw data to actionable predictions. 

Scene 14

The learning pipeline from raw data to actionable predictions

SECTION 4 · HOW ML WORKS. The Machine Learning Pipeline. 

Scene 15

Each step is essential. Skipping any step leads to poor model performance.

SECTION 4 · HOW ML WORKS. Training vs Prediction. 

Scene 16

Feed thousands of past house records to the algorithm

Algorithm analyzes size, location, rooms, age vs actual prices

Discovers mathematical relationships between features and price

Produces a trained model that captures those relationships

SECTION 5. Types of Machine Learning. Supervised · Unsupervised · Reinforcement Learning. 

Scene 17

Supervised · Unsupervised · Reinforcement Learning

SECTION 5 · TYPES OF ML. Supervised Learning. Data + Correct Answers → Learn to Predict New Cases. 

Scene 18

Data + Correct Answers  →  Learn to Predict New Cases

Output: A label (e.g., Churn / Not Churn)

SECTION 5 · TYPES OF ML. Unsupervised Learning. Data without labels → Discover hidden patterns on their own. 

Scene 19

Data without labels  →  Discover hidden patterns on their own

Group similar data points together without predefined categories

No labels needed — model finds groups naturally

Simplify data by reducing number of features while preserving information

SECTION 5 · TYPES OF ML. Reinforcement Learning. AGENT. 

Scene 20

Concept for now — we study RL in depth later in the course.

SECTION 5 · TYPES OF ML. Comparing the Three Types of ML. 

Scene 21

Type Data Requirement Goal Examples Supervised Labeled data (features + target) Predict output for new data Spam detection, price prediction Unsupervised Unlabeled data (features only) Discover hidden patterns/groups Customer segments, clustering Reinforcement Environment + reward signal Learn optimal actions over time Game AI, robotics, self-driving

SECTION 6. The Machine Learning Workflow. An 8-step end-to-end pipeline from problem to deployment. 

Scene 22

An 8-step end-to-end pipeline from problem to deployment

SECTION 6 · ML WORKFLOW. ML Workflow: Steps 1 & 2. 

Scene 23

Before touching any data, clearly define what you want to predict or achieve.
 Example: "Can we predict whether a customer will leave the company next month?"
 Key questions: • What is the business goal? • What does success look like? • Is this regression or classification?

Gather all potentially relevant data for your problem.
 For churn prediction: • Customer age & demographics • Contract type & duration • Monthly payment amount • Usage frequency & patterns • Customer service calls • Previous churn history

SECTION 6 · ML WORKFLOW. ML Workflow: Steps 3 & 4. 

Scene 24

Raw data is almost never ready to use. Cleaning includes:
 • Handle missing values • Remove duplicates • Fix or remove outliers • Encode categorical data (text → numbers) • Scale numeric features
 Garbage in → Garbage out!

Divide your dataset into three parts:
 • Training Set (~70%) — used to train the model • Validation Set (~15%) — tune hyperparameters • Test Set (~15%) — final evaluation only
 Never use test data during training — that's cheating!

SECTION 6 · ML WORKFLOW. ML Workflow: Steps 5 & 6. 

Scene 25

Choose the appropriate ML algorithm for your problem type.
 Common algorithms: • Linear Regression (continuous output) • Logistic Regression (binary classification) • Decision Tree (interpretable rules) • Random Forest (ensemble method) • Support Vector Machine (SVM)

Feed the training data to the algorithm. The algorithm:
 • Learns relationships between features and labels • Adjusts its internal parameters iteratively • Minimizes prediction error over many iterations
 This is the core "learning" step!

SECTION 6 · ML WORKFLOW. ML Workflow: Steps 7 & 8. 

Scene 26

Measure how well the model performs on the test set.
 Key metrics (details in future chapters): • Accuracy — percentage correct • Precision — of predicted positives, how many correct? • Recall — of actual positives, how many found? • Error rate — proportion of mistakes

If performance is not good enough, iterate and improve:
 • Engineer better features • Collect more / better data • Try a different algorithm • Adjust hyperparameters (tuning)
 ML is an iterative process, not a one-shot task!

SECTION 7. ML Applications Across Industries. Business, Finance, Healthcare, E-Commerce & more. 

PptxGenJS Presentation

Scene 1 (0s)

Scene 2 (10s)

Scene 3 (39s)

Scene 4 (48s)

Scene 5 (1m 9s)

Scene 6 (1m 34s)

Scene 7 (1m 56s)

Scene 8 (2m 5s)

Scene 9 (2m 33s)

Scene 10 (2m 50s)

Scene 11 (2m 58s)

Scene 12 (3m 18s)

Scene 13 (3m 55s)

Scene 14 (4m 12s)

Scene 15 (4m 21s)

Scene 16 (4m 42s)

Scene 17 (5m 8s)

Scene 18 (5m 16s)

Scene 19 (5m 41s)

Scene 20 (6m 4s)

Scene 21 (6m 25s)

Scene 22 (6m 42s)

Scene 23 (6m 51s)

Scene 24 (7m 18s)

Scene 25 (7m 42s)

Scene 26 (8m 7s)

Scene 27 (8m 33s)

Scene 28 (8m 42s)

Scene 29 (8m 56s)