Intriguing Properties of Neural Networks - Spring 2021

Explore intriguing properties of neural networks in this 29-minute lecture from the University of Central Florida. Delve into key concepts including adversarial attacks, transfer learning, training overlap, and spectral analysis of instability. Examine semantic information in neural network units and identify potential weaknesses in these systems. Gain a deeper understanding of neural network behavior and challenges through this comprehensive overview of their fascinating characteristics.

Intro
Terminology
Intriguing properties
adversarial attack
transfer
training overlap
spectral analysis of unstability
Semantic information in units
Weaknesses