Skip to main content

Basic Concepts of Deep Learning

This guide introduces the fundamental concepts of deep learning that you'll need to understand while working with TensorWeaver.

What is Deep Learning?​

Deep learning is a subset of machine learning that uses neural networks with multiple layers to learn from data. Unlike traditional programming where we explicitly write rules, deep learning systems learn these rules from examples.

Core Components​

Neurons​

  • The basic computational unit of a neural network
  • Takes multiple inputs, applies weights and bias, and produces an output
  • Similar to biological neurons in concept

Layers​

  • Collections of neurons that process information together
  • Common types:
    • Input Layer: Receives raw data
    • Hidden Layers: Process intermediate representations
    • Output Layer: Produces final predictions

Activation Functions​

  • Non-linear functions that help networks learn complex patterns
  • Common examples:
    • ReLU (Rectified Linear Unit)
    • Sigmoid
    • Tanh

The Learning Process​

Forward Propagation​

  • How data flows through the network
  • Each layer processes the data and passes it to the next
  • Results in a prediction or output

Backward Propagation​

  • How the network learns from its mistakes
  • Calculates gradients to adjust weights and biases
  • Uses the chain rule from calculus

Training Components​

  • Loss Function: Measures how wrong the predictions are
  • Optimizer: Updates weights to minimize the loss
  • Batch Processing: Training on small sets of data at a time