Adversarial Examples of Deep Learning

Paper

Intriguing properties of neural networks

• arXiv: http://arxiv.org/abs/1312.6199
• my notes: In each layer of a deep network it is the “direction” of “space” (ensemble of feature activations) which encodes useful class information rather than individual units (feature activations).

Generative Adversarial Networks

• arXiv: http://arxiv.org/abs/1406.2661
• github: https://github.com/goodfeli/adversarial
• github: https://github.com/aleju/cat-generator

Generative Adversarial Autoencoders in Theano

• blog: https://swarbrickjones.wordpress.com/2016/01/24/generative-adversarial-autoencoders-in-theano/
• github: https://github.com/mikesj-public/dcgan-autoencoder

Deep Neural Networks are Easily Fooled: High Confidence Predictions for Unrecognizable Images

• paper: http://arxiv.org/abs/1412.1897
• code: https://github.com/Evolving-AI-Lab/fooling/

Explaining and Harnessing Adversarial Examples

• introduction: primary cause of neural networks’ vulnerability to adversarial perturbation is their linear nature
• paper: http://arxiv.org/abs/1412.6572

Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks(NIPS 2015)

• arXiv: http://arxiv.org/abs/1506.05751
• code: http://soumith.ch/eyescream/
• project page: http://soumith.ch/eyescream/
• homepage: http://www.cs.nyu.edu/~denton/
• notes: http://colinraffel.com/wiki/deep_generative_image_models_using_a_laplacian_pyramid_of_adversarial_networks

Conditional generative adversarial nets for convolutional face generation

• paper: http://www.foldl.me/uploads/2015/conditional-gans-face-generation/paper.pdf
• blog: http://www.foldl.me/2015/conditional-gans-face-generation/
• github: https://github.com/hans/adversarial

Distributional Smoothing with Virtual Adversarial Training

• paper: http://arxiv.org/abs/1507.00677
• code: https://github.com/takerum/vat

Confusing Deep Convolution Networks by Relabelling

• arXiv: http://arxiv.org/abs/1510.06925v1

Exploring the Space of Adversarial Images

• arXiv: http://arxiv.org/abs/1510.05328
• github: https://github.com/tabacof/adversarial

Learning with a Strong Adversary

• arXiv: http://arxiv.org/abs/1511.03034

Adversarial Autoencoders

• arXiv: http://arxiv.org/abs/1511.05644
• slides: https://docs.google.com/presentation/d/1Lyp91JOSzXo0Kk8gPdgyQUDuqLV_PnSzJh7i5c8ZKjs/edit?pref=2&pli=1
• notes(by Dustin Tran): http://dustintran.com/blog/adversarial-autoencoders/
• TFD manifold: http://www.comm.utoronto.ca/~makhzani/adv_ae/tfd.gif
• SVHN style manifold: http://www.comm.utoronto.ca/~makhzani/adv_ae/svhn.gif

Understanding Adversarial Training: Increasing Local Stability of Neural Nets through Robust Optimization

• arXiv: http://arxiv.org/abs/1511.05432
• github: https://github.com/yutaroyamada/RobustTraining

Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks (DCGAN)

• arxiv: http://arxiv.org/abs/1511.06434
• github: https://github.com/Newmu/dcgan_code
• github: https://github.com/mattya/chainer-DCGAN
• github: https://github.com/soumith/dcgan.torch
• github: https://github.com/carpedm20/DCGAN-tensorflow

Blog

(Deep Learning’s Deep Flaws)’s Deep Flaws (By Zachary Chase Lipton)

http://www.kdnuggets.com/2015/01/deep-learning-flaws-universal-machine-learning.html

Deep Learning Adversarial Examples – Clarifying Misconceptions (By Ian Goodfellow (Google))

http://www.kdnuggets.com/2015/07/deep-learning-adversarial-examples-misconceptions.html

Torch convolutional GAN: Generating Faces with Torch

• blog: http://torch.ch/blog/2015/11/13/gan.html
• github: https://github.com/skaae/torch-gan

Adversarial Machines: Fooling A.Is (and turn everyone into a Manga)

https://medium.com/@samim/adversarial-machines-998d8362e996#.iv3muefgt

How to trick a neural network into thinking a panda is a vulture

https://codewords.recurse.com/issues/five/why-do-neural-networks-think-a-panda-is-a-vulture

Code

Generate cat images with neural networks

• github: https://github.com/aleju/cat-generator