Machine Learning vs Deep Learning
AI, machine learning and deep learning are related words, but they do not mean the same thing.
Three Circles
Think of three circles, one inside the other:
- AI is the biggest circle: any technique that makes computers act smart, including hand-written rules.
- Machine learning (ML) is inside AI: systems that learn patterns from data.
- Deep learning (DL) is inside ML: machine learning that uses large neural networks with many layers.
So every deep learning system is machine learning, and every machine learning system is AI, but not the other way round.
What is Machine Learning?
In machine learning, you give the computer examples (data) and the answers you want, and it finds the rule that connects them. Later it uses that rule on new data.
For example, an online shop in Karachi wants to predict delivery time from distance. It has past orders:
Example
from sklearn.linear_model import LinearRegression
distance_km = [[2], [5], [8], [12]]
minutes = [20, 32, 44, 60]
model = LinearRegression()
model.fit(distance_km, minutes)
print(round(model.predict([[10]])[0]))52
The model learned that each extra kilometre adds about 4 minutes, and it predicts about 52 minutes for 10 km. Nobody wrote that rule by hand.
Types of Machine Learning
| Type | How it learns | Example |
|---|---|---|
| Supervised learning | From examples with correct answers (labels) | Predict house price, detect spam |
| Unsupervised learning | Finds groups or patterns without labels | Group customers by shopping habits |
| Reinforcement learning | Learns by trial and error with rewards | Game-playing AI, robot control |
Modern chatbots use a mix: they first learn from huge amounts of text (a form of self-supervised learning), then are improved with human feedback.
What is Deep Learning?
Deep learning uses neural networks with many layers. Each layer learns a slightly more complex pattern. In a photo model, early layers find edges, middle layers find shapes like eyes and wheels, and later layers recognise whole objects like "cat" or "rickshaw".
Here is a tiny neural network in scikit-learn. It is small, but it uses the same idea:
Example
from sklearn.neural_network import MLPClassifier
# [hours studied, hours slept] -> passed (1) or failed (0)
X = [[1, 4], [2, 5], [6, 7], [8, 8], [3, 4], [7, 6]]
y = [0, 0, 1, 1, 0, 1]
net = MLPClassifier(hidden_layer_sizes=(8, 8), max_iter=2000, random_state=1)
net.fit(X, y)
print(net.predict([[7, 7], [1, 5]]))[1 0]
Real deep learning models have millions or billions of numbers inside them and are trained on GPUs.
Which One to Use?
| Classic machine learning | Deep learning | |
|---|---|---|
| Data needed | Hundreds to thousands of rows | Usually very large datasets |
| Computer power | A normal laptop | Often GPUs |
| Best for | Tables of numbers (sales, marks, prices) | Images, audio, text, video |
| Easy to explain? | Often yes | Usually harder |
For a spreadsheet of sales or student results, simple machine learning is often faster, cheaper and just as accurate as deep learning. Use deep learning when the data is images, sound or language.
Exercise
Exercise
Which statement is correct?