Deep learning uses artificial neural networks with many layers to learn representations from data. Each layer transforms its input into a more abstract representation, allowing the system to capture complex patterns without hand-engineered features.
It requires substantially more data and computation than earlier machine learning approaches.
Alternative Names:
Neural Networks, Deep Neural Networks
Why it Matters?
Deep learning is what made current legal AI capabilities possible, but the layered architecture is also why these systems resist explanation. Unlike a decision tree whose logic can be traced, a deep network's reasoning is distributed across millions of parameters. That opacity is the technical reason source citation matters more than model interpretability for legal use: you verify the output against documents rather than auditing the reasoning.
Frequently Confused with
Related terms
Frequently asked questions
How does deep learning differ from other machine learning?
Why does this matter for legal use?
