Verification of Neural Networks (Lecture Notes)

TL;DR

This paper provides an introduction to neural network verification, covering various architectures, specification languages, and verification algorithms from a theoretical standpoint.

Contribution

It offers a comprehensive theoretical overview of neural network verification methods across multiple neural network architectures.

Findings

Discusses verification techniques for feed-forward and recurrent neural networks.

Includes analysis of attention mechanisms and transformers.

Introduces specification languages for neural network verification.

Abstract

These lecture notes provide an introduction to the verification of neural networks from a theoretical perspective. We discuss feed-forward neural networks, recurrent neural networks, attention mechanisms, and transformers, together with specification languages and algorithmic verification techniques.