What is AI?
Definition:
AI (Artificial Intelligence) is the science of creating machines or software that can think,
learn, and make decisions like humans.
Purpose / logic:
● Simulate human intelligence
● Automate tasks requiring reasoning & perception
● Solve complex problems faster than humans
Example:
Virtual assistants, language translators, self-driving cars.
2. Abbreviations (must-know)
Abbreviation Full Form
AI Artificial Intelligence
ML Machine Learning
DL Deep Learning
NLP Natural Language Processing
CNN Convolutional Neural Network
RNN Recurrent Neural Network
ANN Artificial Neural Network
RL Reinforcement Learning
KNN K-Nearest Neighbors
SVM Support Vector Machine
3. AI: Types (based on capability)
Type Logic Example
Narrow AI Task-specific Siri, spam filter
General AI Human-like Theoretical
Super AI Beyond Not yet exists
human
4. AI: Types (based on functionality)
Type Logic Example
Reactive No memory Chess-playing AI
Limited memory Uses past data Self-driving car
Theory of mind Understand Theoretical
emotions
Self-aware Has consciousness Not yet exists
5. Machine Learning (ML)
Definition:
ML is a subset of AI that uses data to train models to make predictions or decisions
without explicit programming.
Purpose:
Learn patterns from data → predict outcomes.
Example:
Spam detection, recommendation systems.
6. Deep Learning (DL)
Definition:
DL is a subset of ML using multi-layered neural networks to learn complex patterns.
Purpose:
Work with big data (images, text, speech).
Example:
Face recognition, voice assistants.
7. Difference: AI vs ML vs DL
AI ML DL
Scope Broad Subset of AI Subset of ML
Human-like Yes Partial Focused
Data need Low–medium High Very high
Example Planning robot Spam filter Face
detection
8. ML: Types
Type Logic Example
Supervised Labeled data Email spam detection
Unsupervised Unlabeled data Customer
segmentation
Semi-supervise Few labels Text classification
d
Reinforcement Learn via reward Game AI, robotics
9. Key ML Algorithms
Algorithm Purpose
Linear Predict continuous value
regression
Decision tree Classification/regression
KNN Classification by neighbors
SVM Classify with hyperplane
K-means Clustering
Naïve Bayes Probabilistic classifier
10. Deep Learning: Neural Networks
Definition:
Models inspired by human brain → layers of neurons.
Network Purpose Example
ANN General Tabular data
tasks
CNN Images Object detection
RNN Sequences Language modeling
11. Natural Language Processing (NLP)
Definition:
AI to understand & generate human language.
Example:
Chatbots, translators, sentiment analysis.
12. Computer Vision (CV)
Definition:
AI for analyzing & understanding images/videos.
Example:
Face unlock, medical imaging.
13. Reinforcement Learning (RL)
Definition:
Agent learns by trial & error using rewards/punishments.
Example:
AlphaGo, robot learning to walk.
14. Overfitting vs Underfitting
Overfitting Underfitting
Meaning Too complex → memorizes Too simple → misses
pattern
Accuracy High train, low test Low train & test
15. Training vs Testing
Training Testing
Use Build model Evaluate
model
Data Seen Unseen
16. Data Preprocessing
Purpose:
Improve data quality.
Step Example
Cleaning Remove
duplicates
Scaling Normalize values
Encoding Text → numbers
17. Activation Functions (DL)
Function Purpose
Sigmoid Output 0–1
ReLU Faster training
Softmax Multi-class output
18. Loss Function & Optimizer
Purpose Example
Loss Measure MSE, cross-entropy
error
Optimizer Reduce loss SGD, Adam
19. Hyperparameters
Definition:
Values set before training (e.g., learning rate, epochs).
Purpose:
Control training behavior.
20. Use Cases
Area Example
Healthcare Disease
prediction
Finance Fraud detection
Retail Recommendation
s
Automotive Self-driving cars
Language Speech-to-text
22. Deep vs Shallow Networks
Shallo Deep
w
Layers Few Many
Task Simple Complex
23. AI Ethics
Concern Example
Bias Skewed
predictions
Privacy Data misuse
Transparenc Explainable AI
y
24. Big Data & AI
Purpose:
AI helps analyze big data; big data fuels AI models.
25. Why study AI & ML?
● Solve real-world problems
● Automate processes
● Build smart systems
● Career demand
To remember easily:
● AI → parent, ML → child, DL → grandchild
● Supervised → labels, Unsupervised → no labels, RL → reward
● CNN → images, RNN → sequences
