Whitening Black-Box Neural Networks

11/06/2017
by   Seong Joon Oh, et al.
0

Many deployed learned models are black boxes: given input, returns output. Internal information about the model, such as the architecture, optimisation procedure, or training data, is not disclosed explicitly as it might contain proprietary information or make the system more vulnerable. This work shows that such attributes of neural networks can be exposed from a sequence of queries. This has multiple implications. On the one hand, our work exposes the vulnerability of black-box neural networks to different types of attacks -- we show that the revealed internal information helps generate more effective adversarial examples against the black box model. On the other hand, this technique can be used for better protection of private content from automatic recognition models using adversarial examples. Our paper suggests that it is actually hard to draw a line between white box and black box models.

READ FULL TEXT

page 5

page 11

page 12

research
11/03/2018

CAAD 2018: Powerful None-Access Black-Box Attack Based on Adversarial Transformation Network

In this paper, we propose an improvement of Adversarial Transformation N...
research
11/15/2018

A note on hyperparameters in black-box adversarial examples

Since Biggio et al. (2013) and Szegedy et al. (2013) first drew attentio...
research
10/18/2022

It's a long way! Layer-wise Relevance Propagation for Echo State Networks applied to Earth System Variability

Artificial neural networks (ANNs) are known to be powerful methods for m...
research
06/23/2018

DALEX: explainers for complex predictive models

Predictive modeling is invaded by elastic, yet complex methods such as n...
research
07/21/2019

Open DNN Box by Power Side-Channel Attack

Deep neural networks are becoming popular and important assets of many A...
research
01/21/2021

Copycat CNN: Are Random Non-Labeled Data Enough to Steal Knowledge from Black-box Models?

Convolutional neural networks have been successful lately enabling compa...
research
11/15/2016

Iterative Orthogonal Feature Projection for Diagnosing Bias in Black-Box Models

Predictive models are increasingly deployed for the purpose of determini...

Please sign up or login with your details

Forgot password? Click here to reset