Representation of White- and Black-Box Adversarial Examples in Deep Neural Networks and Humans: A Functional Magnetic Resonance Imaging Study

Han, Chihye; Yoon, Wonjun; Kwon, Gihyun; Nam, Seungkyu; Kim, Daeshik

doi:10.48550/arxiv.1905.02422

Cited by 1 publication

(1 citation statement)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Neural networks performance is often compared to human performance (Baker et al, 2018;Han et al, 2019;Srivastava et al, 2019;Ma and Peters, 2020). A fundamental feature of human perception is that it supports widespread generalization, including combinatorial generalization (e.g., identifying and understanding images composed of novel combination of known features).…”

Section: Discussionmentioning

confidence: 99%

Convolutional Neural Networks Are Not Invariant to Translation, but They Can Learn to Be

Biscione,

Bowers

2021

Preprint

View full text Add to dashboard Cite

When seeing a new object, humans can immediately recognize it across different retinal locations: the internal object representation is invariant to translation. It is commonly believed that Convolutional Neural Networks (CNNs) are architecturally invariant to translation thanks to the convolution and/or pooling operations they are endowed with. In fact, several studies have found that these networks systematically fail to recognise new objects on untrained locations. In this work, we test a wide variety of CNNs architectures showing how, apart from DenseNet-121, none of the models tested was architecturally invariant to translation. Nevertheless, all of them could learn to be invariant to translation. We show how this can be achieved by pretraining on ImageNet, and it is sometimes possible with much simpler data sets when all the items are fully translated across the input canvas. At the same time, this invariance can be disrupted by further training due to catastrophic forgetting/interference. These experiments show how pretraining a network on an environment with the right 'latent' characteristics (a more naturalistic environment) can result in the network learning deep perceptual rules which would dramatically improve subsequent generalization.

show abstract

Section: Discussionmentioning

confidence: 99%