🧠 Deep Learning

Neural Network Builder

Add layers, tune neurons, change activation functions — then watch forward propagation light up your network with real computed values at every node!

🏗️ Build Network

⚡ Forward Pass

🔬 Activations

📉 Training

🏆 Badge

How Neural Networks Work

⚪

Neurons

Each neuron receives inputs, multiplies by weights, adds a bias, then passes through an activation function.

🔗

Connections

Every neuron in one layer connects to every neuron in the next layer. Each connection has a learnable weight.

➡️

Forward Pass

Data flows left to right through the network, transforming at each layer, until it reaches the output.

🔄

Backpropagation

Error flows backwards from output to input, adjusting weights to reduce mistakes. Repeat thousands of times!

🧠

Wizzy the AI Tutor

Build your neural network! 🏗️ Use the controls on the right to add/remove layers and change the number of neurons in each layer. The network starts with 2 inputs, one hidden layer, and 1 output. Watch the canvas update live!

Step 1 — Design Your Architecture

2 → 3 → 1

🧠

Wizzy the AI Tutor

Watch the forward pass! 🔥 Set the input values with the sliders, then press "Run Forward Pass". Each neuron lights up showing its computed value. The animation travels left to right through every layer. Can you spot how the values transform?

Step 2 — Forward Propagation

🎚️ Set Input Values

📊 Layer Outputs

// Press "Run Forward Pass" to start

🧠

Wizzy the AI Tutor

Activation functions are the secret sauce of neural networks! Without them, no matter how many layers you add, it's just linear maths. ReLU, Sigmoid, and Tanh each have different shapes — explore them and see how your input transforms through each one!

Step 3 — Activation Functions Explorer

Try input value: x = 0.5

Drag the slider above to see how each activation function transforms the input value!

🧠

Wizzy the AI Tutor

Watch the network learn to solve XOR — a classic problem! XOR: inputs (0,0)→0, (1,0)→1, (0,1)→1, (1,1)→0. A single neuron can't do this — you need hidden layers! Watch the loss decrease as the network discovers the pattern through backpropagation!

Step 4 — Watch the Network Learn XOR

📉 Loss over Epochs

Epoch: 0

Loss: —

LR: 0.1

Status: Ready

XOR Truth Table — Current Predictions

Press Start Training to begin! The network will adjust its weights using backpropagation.

🧠

Wizzy the AI Tutor

🎊 You've built a neural network, run forward propagation, explored activation functions, and watched backpropagation train a network on XOR! You understand the core of deep learning — the same foundations behind ChatGPT, AlphaGo, and image recognition!

🧠

Deep Learning Badge!

You built a neural network and watched it learn through backpropagation!

Optional. Stays on this device only — not sent to WhizzStep.

🧠 WhizzStep AI Lab

Activity completion card for

Student Name

has built and trained a Neural Network from scratch

This records completion of a browser activity only. It is not an accredited certificate or proof of mastery.

Deep Learning Expert

Backprop Master

Neural Architect

Architecture: — · whizzstep.in

Key Concepts Mastered

Weights & Biases

⚖️ The Learnable Params

Each connection has a weight; each neuron has a bias. Training adjusts all of these to minimise the error.

Activation Function

🔌 Non-Linearity

ReLU, Sigmoid, Tanh — without these, a 1000-layer network is mathematically the same as 1 linear layer.

Forward Pass

➡️ Input to Output

Data multiplied by weights, summed, activated — repeated layer by layer until the output neuron.

Loss Function

📏 Measuring Error

Mean Squared Error (MSE) quantifies how wrong the output is. Training minimises this number.

Backpropagation

🔄 Learning Rule

Using the chain rule of calculus to compute how much each weight contributed to the error, then adjusting it.

Gradient Descent

⛰️ Finding the Valley

Take small steps in the direction of steepest descent on the loss surface. Learning rate controls the step size.

About this lab

AI-04 Data and Coding Classes 9-12 20 min Applied

Learning objective: Assemble a small neural network by hand and watch it adjust its weights as it learns a simple pattern.

What this simplifies: A tiny network with a handful of neurons is used; production neural networks have millions to billions of parameters.

Privacy: No learner input leaves the device.

Teacher prompt: Ask the class why this simulation might mislead someone who takes it too literally.

Reflect: What is one thing this activity showed you that you did not expect?

← Fair AI or Biased AI? Back to All Labs Inside the Black Box →

Bring this learning to my school →