filmov
tv
Adversarial Examples, Optical Illusions and Neural Networks
Показать описание
In this video I look into how researchers discovered AI illusions. I explain how adversarial examples work in neural networks and how we can use them to open up the black box of the neural networks. We discover that adversarial examples are as powerful as optical illusions are in humans to better understand how human perception works.
All references mentioned in the video can be found here:
All references mentioned in the video can be found here:
Adversarial Examples, Optical Illusions and Neural Networks
Adversarial Attacks on Neural Networks - Bug or Feature?
Tricking Artificial Intelligence by Adversarial examples | Patch (Google)
Adversarial examples for humans
Optical Illusions Explained
Adversarial Examples
Adversarial Examples and Human-ML Alignment
Lecture 16 | Adversarial Examples and Adversarial Training
What Are Adversarial Examples?
Adversarial Examples in Deep Learning
Adaptive-Gravity: A Defense Against Adversarial Samples
Adversarial Examples - AI Seminars
Adversarial Geometry-Aware Human Motion Prediction
'How neural networks learn' - Part II: Adversarial Examples
Adversarial Examples: instability in deep learning or vision in general? - by George Cevora
News Techcology - Google ‘optical illusion’ stickers make AI hallucinate
CS 182: Lecture 20: Part 2: Adversarial Examples
Sebastien Bubeck (MSR) -- Adversarial Examples from Computational Constraints
Defense Against the Dark Arts
How Jailbreakers Try to “Free” AI
Why Large Language Models Hallucinate
AI and Society: 12d. Adversarial attacks on neural networks
The Anti 3D Glasses
Nicholas Carlini: Making and Measuring Progress in Adversarial Machine Learning
Комментарии