BUILDING YOUR FIRST NEURAL NETWORKS

In this chapter, you will learn how to build your very first neural networks. These networks are the foundation of Deep Learning and help computers recognize patterns—just like human learning. You will explore the Single Layer Perceptron, the Multi-Layer Perceptron (MLP), and learn how to solve simple tasks like classification and regression using neural networks.

1. Single Layer Perceptron (SLP)

The Single Layer Perceptron is the simplest type of neural network. It has:

One input layer
One output layer
No hidden layers

It is mostly used for simple tasks where the data can be separated into two classes.

How It Works

A perceptron:

Takes input values
Multiplies them by weights
Adds a bias
Uses an activation function
Produces an output (0 or 1)

Real-World Example

A perceptron can help classify:

Whether an email is spam or not
Whether a student passed or failed
Whether a fruit is ripe or unripe

Simple Diagram (Conceptual)

Inputs → [Perceptron] → Output

Code Example: Single Layer Perceptron (Keras)

import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense

# Single Layer Perceptron
model = Sequential([
    Dense(1, activation='sigmoid', input_shape=(2,))
])

model.compile(
    optimizer='sgd',
    loss='binary_crossentropy',
    metrics=['accuracy']
)

print(model.summary())

2. Multi-Layer Perceptron (MLP)

A Multi-Layer Perceptron has more than one layer.

It includes:

Input layer
One or more hidden layers
Output layer

Each hidden layer helps the network learn deeper and more complex patterns.

Why MLPs Are Powerful

Can learn curved boundaries
Can identify shapes in data
Can solve problems SLP cannot solve

Example of Hidden Layer Magic

If we want to classify images of handwritten numbers (0–9), an MLP:

First hidden layer detects simple edges
Second layer detects shapes
Third layer identifies the actual number

Code Example: Simple MLP for Classification

import tensorflow as tf
from tensorflow.keras import Sequential
from tensorflow.keras.layers import Dense

# Multi-Layer Perceptron
model = Sequential([
    Dense(16, activation='relu', input_shape=(4,)),
    Dense(8, activation='relu'),
    Dense(3, activation='softmax')
])

model.compile(
    optimizer='adam',
    loss='sparse_categorical_crossentropy',
    metrics=['accuracy']
)

print(model.summary())

3. Simple Classification Tasks

Classification means predicting a category.

Examples of classification:

Predicting whether a picture contains a cat or dog
Predicting whether a student will pass or fail
Predicting weather type (sunny, rainy, cloudy)

How Classification Works

Input data → Neural Network
Network learns patterns
Output → probability of each class

Code Example: Classification with MLP

import numpy as np
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense

# Fake data (for demonstration)
X = np.random.rand(200, 4)      # 200 samples, 4 input features
y = np.random.randint(0, 2, 200) # Binary labels: 0 or 1

model = Sequential([
    Dense(8, activation='relu', input_shape=(4,)),
    Dense(1, activation='sigmoid')
])

model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])

model.fit(X, y, epochs=5, batch_size=16)

4. Simple Regression Tasks

Regression predicts numbers, not categories.

Examples of Regression

Predicting house prices
Predicting marks based on study hours
Predicting temperature

How Regression Works

The neural network outputs a continuous value
No softmax or sigmoid needed
Usually uses a linear output layer

Code Example: Regression with MLP

import numpy as np
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense

# Fake data for regression
X = np.random.rand(300, 3)       # 300 samples, 3 features
y = X[:, 0] * 5 + 2              # A simple relationship for demonstration

model = Sequential([
    Dense(8, activation='relu', input_shape=(3,)),
    Dense(1)   # Linear output for regression
])

model.compile(optimizer='adam', loss='mse')

model.fit(X, y, epochs=5, batch_size=16)

5. Understanding the Difference: Classification vs Regression

Feature	Classification	Regression
Output type	Category/label	Continuous number
Activation	Sigmoid / Softmax	Linear
Loss function	Cross-entropy	MSE
Example	Cat vs Dog	Predict house price

6. Real-World Examples of Neural Networks

A. Single Layer Perceptron

Used for:

Yes/No questions
Basic sensor decisions
Spam filtering

B. Multi-Layer Perceptron

Used for:

Handwriting recognition
Predicting marks
Simple image classification

C. Classification Tasks

Used for:

Medical disease detection
Student performance prediction
Fraud detection

D. Regression Tasks

Used for:

Predicting salaries
Energy consumption forecasting
Predicting crop yield

BUILDING YOUR FIRST NEURAL NETWORKS

1. Single Layer Perceptron (SLP)

The Single Layer Perceptron is the simplest type of neural network. It has:

One input layer
One output layer
No hidden layers

It is mostly used for simple tasks where the data can be separated into two classes.

How It Works

A perceptron:

Takes input values
Multiplies them by weights
Adds a bias
Uses an activation function
Produces an output (0 or 1)

Real-World Example

A perceptron can help classify:

Whether an email is spam or not
Whether a student passed or failed
Whether a fruit is ripe or unripe

Simple Diagram (Conceptual)

Inputs → [Perceptron] → Output

Code Example: Single Layer Perceptron (Keras)

import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense

# Single Layer Perceptron
model = Sequential([
    Dense(1, activation='sigmoid', input_shape=(2,))
])

model.compile(
    optimizer='sgd',
    loss='binary_crossentropy',
    metrics=['accuracy']
)

print(model.summary())

2. Multi-Layer Perceptron (MLP)

A Multi-Layer Perceptron has more than one layer.

It includes:

Input layer
One or more hidden layers
Output layer

Each hidden layer helps the network learn deeper and more complex patterns.

Why MLPs Are Powerful

Can learn curved boundaries
Can identify shapes in data
Can solve problems SLP cannot solve

Example of Hidden Layer Magic

If we want to classify images of handwritten numbers (0–9), an MLP:

First hidden layer detects simple edges
Second layer detects shapes
Third layer identifies the actual number

Code Example: Simple MLP for Classification

import tensorflow as tf
from tensorflow.keras import Sequential
from tensorflow.keras.layers import Dense

# Multi-Layer Perceptron
model = Sequential([
    Dense(16, activation='relu', input_shape=(4,)),
    Dense(8, activation='relu'),
    Dense(3, activation='softmax')
])

model.compile(
    optimizer='adam',
    loss='sparse_categorical_crossentropy',
    metrics=['accuracy']
)

print(model.summary())

3. Simple Classification Tasks

Classification means predicting a category.

Examples of classification:

Predicting whether a picture contains a cat or dog
Predicting whether a student will pass or fail
Predicting weather type (sunny, rainy, cloudy)

How Classification Works

Input data → Neural Network
Network learns patterns
Output → probability of each class

Code Example: Classification with MLP

import numpy as np
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense

# Fake data (for demonstration)
X = np.random.rand(200, 4)      # 200 samples, 4 input features
y = np.random.randint(0, 2, 200) # Binary labels: 0 or 1

model = Sequential([
    Dense(8, activation='relu', input_shape=(4,)),
    Dense(1, activation='sigmoid')
])

model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])

model.fit(X, y, epochs=5, batch_size=16)

4. Simple Regression Tasks

Regression predicts numbers, not categories.

Examples of Regression

Predicting house prices
Predicting marks based on study hours
Predicting temperature

How Regression Works

The neural network outputs a continuous value
No softmax or sigmoid needed
Usually uses a linear output layer

Code Example: Regression with MLP

import numpy as np
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense

# Fake data for regression
X = np.random.rand(300, 3)       # 300 samples, 3 features
y = X[:, 0] * 5 + 2              # A simple relationship for demonstration

model = Sequential([
    Dense(8, activation='relu', input_shape=(3,)),
    Dense(1)   # Linear output for regression
])

model.compile(optimizer='adam', loss='mse')

model.fit(X, y, epochs=5, batch_size=16)

5. Understanding the Difference: Classification vs Regression

Feature	Classification	Regression
Output type	Category/label	Continuous number
Activation	Sigmoid / Softmax	Linear
Loss function	Cross-entropy	MSE
Example	Cat vs Dog	Predict house price

6. Real-World Examples of Neural Networks

A. Single Layer Perceptron

Used for:

Yes/No questions
Basic sensor decisions
Spam filtering

B. Multi-Layer Perceptron

Used for:

Handwriting recognition
Predicting marks
Simple image classification

C. Classification Tasks

Used for:

Medical disease detection
Student performance prediction
Fraud detection

D. Regression Tasks

Used for:

Predicting salaries
Energy consumption forecasting
Predicting crop yield

deep-learning Topics

deep-learning Tutorial

BUILDING YOUR FIRST NEURAL NETWORKS

1. Single Layer Perceptron (SLP)

How It Works

Real-World Example

Simple Diagram (Conceptual)

Code Example: Single Layer Perceptron (Keras)

2. Multi-Layer Perceptron (MLP)

Why MLPs Are Powerful

Example of Hidden Layer Magic

Code Example: Simple MLP for Classification

3. Simple Classification Tasks

How Classification Works

Code Example: Classification with MLP

4. Simple Regression Tasks

Examples of Regression

How Regression Works

Code Example: Regression with MLP

5. Understanding the Difference: Classification vs Regression

6. Real-World Examples of Neural Networks

A. Single Layer Perceptron

B. Multi-Layer Perceptron

C. Classification Tasks

D. Regression Tasks

deep-learning Topics

deep-learning Tutorial

BUILDING YOUR FIRST NEURAL NETWORKS

1. Single Layer Perceptron (SLP)

How It Works

Real-World Example

Simple Diagram (Conceptual)

Code Example: Single Layer Perceptron (Keras)

2. Multi-Layer Perceptron (MLP)

Why MLPs Are Powerful

Example of Hidden Layer Magic

Code Example: Simple MLP for Classification

3. Simple Classification Tasks

How Classification Works

Code Example: Classification with MLP

4. Simple Regression Tasks

Examples of Regression

How Regression Works

Code Example: Regression with MLP

5. Understanding the Difference: Classification vs Regression

6. Real-World Examples of Neural Networks

A. Single Layer Perceptron

B. Multi-Layer Perceptron

C. Classification Tasks

D. Regression Tasks