Chapter 2: Basics of Neural Networks

🎯 Objective

By the end of this chapter, learners will:

• Understand what a neural network is and why it’s useful
• Identify the parts of a basic neural network (neurons, layers, weights, etc.)
• Visualize how data flows through a network
• Explore how neural networks “learn” from data

🤔 What Is a Neural Network?

A neural network is a computer system inspired by the human brain. It’s made up of layers of nodes (neurons) that work together to make decisions—like recognizing a cat in a photo or translating a sentence.

🧠 Real-Life Analogy:

Imagine a brain made of LEGOs. Each block (neuron) connects to another, passing info along until a decision pops out.

🧱 The Basic Building Blocks

🎯 Simple Diagram: Feedforward Neural Network

  Input       Hidden        Output
[  x1  ] → [   ●   ●   ] → [   ●   ]
[  x2  ] → [   ●   ●   ] → [   ●   ]

🧠 Each layer transforms the input just a little bit more—until the final decision is made.

🎮 Activity 1: Neural Network Flow Game

Let’s pretend to be a neural network!

1. Inputs: “It’s raining” (x1), “Have umbrella” (x2)
2. You are the hidden layer:
   • Multiply x1 by weight 0.6
   • Multiply x2 by weight 0.9
3. Add both values
4. Apply the rule:
   • If result > 0.75 → Output: “Go outside”
   • Else → “Stay inside”

📓 Try it with:

x1 = 1  # raining
x2 = 0  # no umbrella
w1 = 0.6
w2 = 0.9

z = (x1 * w1) + (x2 * w2)
print("Should you go outside?", "Yes" if z > 0.75 else "No")

🔁 How Do Neural Networks Learn?

🧠 They learn by trial and error—just like humans!

📊 Training Steps:

1. Feed Data → Pass real examples into the network
2. Predict Output → See what the network guesses
3. Check Error → Compare guess with correct answer
4. Adjust Weights → Use a process called backpropagation
5. Repeat… a lot.

💥 Fun Fact

The idea of neural networks has existed since the 1950s—but thanks to today’s computing power, they’re now able to solve problems that were once impossible.

🧪 Activity 2: Build a Mini Neural Net in TensorFlow

Try this in Google Colab:

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

# Define a simple model
model = Sequential([
    Dense(4, activation='relu', input_shape=(2,)),
    Dense(1, activation='sigmoid')
])

# Compile the model
model.compile(optimizer='adam', loss='binary_crossentropy')

print("Neural network ready!")

🧠 This builds a simple neural network with:

• 2 input features
• 1 hidden layer with 4 neurons
• 1 output node for yes/no prediction

💬 Reflection Prompt

“If your brain were a neural network, what kind of problems would it solve best?”

Write or sketch your answer. You can imagine a “dream AI” that sorts music, gives advice, or wins at chess.

🧠 Key Takeaways

✅ Neural networks are modeled after the human brain
✅ They use layers of neurons to transform and analyze data
✅ Learning happens by adjusting weights over time
✅ Neural networks are the heart of AI—from chatbots to self-driving cars

🚀 Next Up:

Chapter 3: Building Your First Neural Network with Real Data
Time to train a real model and see it learn—one example at a time!