Table of Contents

Artificial Neural Networks

Rumman Ansari June 26, 2026 14 views Subject Details

DEEP LEARNING

Artificial Neural Networks (ANN)

The foundation of Deep Learning — inspired by the human brain to solve complex problems.

What is an Artificial Neural Network?

An Artificial Neural Network (ANN) is a computational model inspired by the way the human brain processes information. It consists of layers of interconnected artificial neurons that learn patterns from data.

In simple words — An ANN is a digital brain that learns from data to make predictions.

Why are ANNs Important?

Form the foundation of all Deep Learning models.
Used in image, speech, and text recognition.
Capable of learning from massive amounts of data.
Automatically extract features without manual coding.
Power modern AI systems like ChatGPT, Tesla, and Siri.

If Deep Learning is the brain of AI, ANNs are its cells.

Inspired by the Human Brain

The structure of an ANN is similar to the biological neuron system in our brain.

Biological Neuron

Dendrites receive input
Cell body processes
Axon sends output

Artificial Neuron

Inputs from previous neurons
Weighted sum + activation
Sends output to next layer

Structure of an Artificial Neural Network

Input Layer

Receives raw input data — like image pixels, text, or numerical values.

Hidden Layers

Process and transform the data using neurons and activation functions.

Output Layer

Produces the final prediction or result.

Key Components of ANN

Neuron

The smallest processing unit of an ANN.

Weights

Numbers that determine the importance of each input.

Bias

An additional parameter that helps adjust the output.

Activation Function

Adds non-linearity, allowing the model to learn complex patterns.

Forward Propagation

Data flows from input → hidden → output to make predictions.

Backpropagation

Algorithm to update weights and reduce error.

Loss Function

Measures how far predictions are from the true output.

Optimizer

Updates weights to minimize the loss (e.g., SGD, Adam).

Mathematics of a Neuron

Each neuron processes inputs using this formula:

NEURON OUTPUT

$$ z = w_1x_1 + w_2x_2 + \dots + w_nx_n + b $$ $$ a = f(z) $$

Where:

x — input values
w — weights
b — bias
f — activation function
a — final neuron output

Common Activation Functions

Function	Formula	Use Case
Sigmoid	1 / (1 + e⁻ᶻ)	Binary classification
Tanh	(eᶻ - e⁻ᶻ) / (eᶻ + e⁻ᶻ)	Hidden layers
ReLU	max(0, z)	Most modern networks
Leaky ReLU	max(0.01z, z)	Avoids dead neurons
Softmax	eᶻᵢ / Σ eᶻ	Multi-class output

How Does an ANN Learn?

Learning Process

Input data is passed into the network.
Each layer transforms the data.
Network produces an output.
Loss function calculates the error.
Backpropagation adjusts weights.
The process repeats until accuracy improves.

Types of Artificial Neural Networks

Feedforward Neural Network (FNN)

Information moves in one direction — input to output.

Recurrent Neural Network (RNN)

Used for sequential data — text, speech, time series.

Convolutional Neural Network (CNN)

Used for image and computer vision tasks.

LSTM & GRU

Specialized RNNs for handling long-term dependencies.

Generative Adversarial Networks (GANs)

Used for image generation, deepfakes, AI art.

Autoencoders

Used for dimensionality reduction and anomaly detection.

Real-Life Analogy

ANN = Brain of a Student

Just like a student learns by adjusting their understanding after every mistake, an ANN adjusts its weights using backpropagation. Over many lessons (epochs), it becomes smarter and more accurate.

Python Example — Simple ANN

Prerequisites: Python 3.x, tensorflow, numpy.

pip install tensorflow numpy

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

# Sample data: predict sum > 1
X = np.array([[0,0], [0,1], [1,0], [1,1]])
y = np.array([0, 1, 1, 1])

# Define ANN
model = Sequential([
    Dense(8, activation="relu", input_shape=(2,)),
    Dense(4, activation="relu"),
    Dense(1, activation="sigmoid")
])

# Compile model
model.compile(optimizer="adam", loss="binary_crossentropy", metrics=["accuracy"])

# Train model
model.fit(X, y, epochs=200, verbose=0)

# Predict
print(model.predict(np.array([[1, 1]])))

Output The ANN learns to predict whether the sum of inputs is greater than 1.

ANN vs Traditional ML

Aspect	ANN	Traditional ML
Feature Engineering	Automatic	Manual
Data Requirement	Large	Small to medium
Hardware	GPU needed	CPU works
Complex Tasks	Excellent	Limited
Performance	Improves with data	Plateaus

Real-World Applications

Computer Vision

Face recognition
Object detection

Speech Recognition

Siri, Alexa
Voice typing

Natural Language Processing

Chatbots
Sentiment analysis

Autonomous Vehicles

Lane detection
Traffic recognition

Healthcare

Cancer detection
X-ray analysis

E-commerce

Personalized recommendations
Customer segmentation

Finance

Fraud detection
Stock predictions

Creative AI

AI Art
Music creation

Advantages of ANN

Learns complex non-linear patterns.
Automatic feature extraction.
Excellent at handling unstructured data.
Improves with more data.
Used across multiple AI fields.

Disadvantages

Limitation 1 Needs huge computational resources.

Limitation 2 Long training time.

Limitation 3 Acts like a black-box (hard to interpret).

Limitation 4 Risk of overfitting on small datasets.

Common Mistakes to Avoid

Mistake 1 Not normalizing or scaling input data.

Mistake 2 Choosing too many layers when not needed.

Mistake 3 Using wrong activation functions.

Mistake 4 Ignoring overfitting by skipping dropout.

Best Practices

Quick Tips

Start with small networks.
Normalize data before training.
Use proper activation functions.
Apply dropout to prevent overfitting.
Tune learning rate carefully.
Always validate using cross validation.

Importance of ANN

Foundation of AI

Core of all DL models
Essential ML skill

Powerful Predictions

Used in modern industries
Drives smart systems

Drives Innovation

Self-driving cars
Generative AI

Career Opportunity

In-demand skill
High-paying jobs

Golden Rule

REMEMBER

Neurons + Layers + Learning = Intelligence

Key Takeaway

Artificial Neural Networks (ANNs) are the building blocks of modern Deep Learning. They mimic the human brain to learn patterns and make smart predictions. From chatbots to self-driving cars, ANN powers the AI systems shaping our future.