AI Dictionary › Fondamenti AI

Activation Function

Funzione di attivazione

The activation function introduces non-linearity into neural networks, deciding whether and how strongly a neuron 'fires' in response to its input. Without it, stacking many layers would equal a single linear transformation, unable to capture complex relationships. It's like an adjustable switch: given an incoming signal, it decides what signal to pass to the next neuron. The most common today is ReLU, which passes positive values and zeroes out negatives, simple and efficient. Others like sigmoid and hyperbolic tangent squash the output into a bounded range, useful in specific cases but prone to vanishing-gradient problems.

Definition

The activation choice affects training speed and a network's learning capacity. Thanks to the non-linearity it adds, a model can approximate highly intricate functions, recognize faces, translate languages, or generate text, tasks impossible for a plain linear system.

Related terms

• Neural Network
• Deep Learning
• Backpropagation
• Softmax
• Parameters

More in Fondamenti AI

• LLM (Large Language Model)
• Token
• Context Window
• Temperature
• AI Hallucination
• Fine-tuning
• RAG (Retrieval Augmented Generation)
• Embedding

Put it into practice

• Real scenario: apply this concept with an instant AI verdict.
• Daily gym: short drills to train dictionary concepts.
• Playoff game: challenge other players, elimination-style.

📱 Download the Android app (beta) iOS coming soon

Say what you mean. Get what you need.

Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.