A Feedforward Neural Network (FNN) is the simplest type of artificial neural network. Information flows in one direction—from input to output—without any loops or cycles. Think of it as a one-way street where data enters, gets processed through hidden layers, and produces a result.
Watch how signals propagate through a feedforward network
The "feedforward" name comes from how information flows: it always moves forward from input to output. Unlike recurrent networks, there are no feedback loops or memory of past inputs. Each prediction is independent and based solely on the current input.
When you upload a photo to identify a dog breed:
Input Layer: Pixel values (784 pixels = 28×28 image)
Hidden Layer 1: 128 neurons detect edges and simple shapes
Hidden Layer 2: 64 neurons recognize patterns like ears, eyes
Output Layer: 10 neurons, one per breed (highest value wins)
1. Forward Pass: Input data flows through the network to generate predictions.
2. Loss Calculation: Compare predictions to actual values using a loss function.
3. Backward Pass (Backpropagation): Calculate gradients to determine how to adjust weights.
4. Weight Update: Use gradient descent to minimize loss.
5. Repeat: Iterate thousands of times until the network learns the patterns.