CNNs are Myopic

05/22/2022
by   Vamshi C. Madala, et al.
0

We claim that Convolutional Neural Networks (CNNs) learn to classify images using only small seemingly unrecognizable tiles. We show experimentally that CNNs trained only using such tiles can match or even surpass the performance of CNNs trained on full images. Conversely, CNNs trained on full images show similar predictions on small tiles. We also propose the first a priori theoretical model for convolutional data sets that seems to explain this behavior. This gives additional support to the long standing suspicion that CNNs do not need to understand the global structure of images to achieve state-of-the-art accuracies. Surprisingly it also suggests that over-fitting is not needed either.

READ FULL TEXT

page 2

page 7

page 8

page 9

page 14

page 15

research
05/23/2019

Interpreting Adversarially Trained Convolutional Neural Networks

We attempt to interpret how adversarially trained convolutional neural n...
research
09/27/2021

Leveraging Multiple CNNs for Triaging Medical Workflow

High hospitalization rates due to the global spread of Covid-19 bring ab...
research
08/04/2022

Estimating relative diffusion from 3D micro-CT images using CNNs

In the past several years, convolutional neural networks (CNNs) have pro...
research
05/17/2020

North Atlantic Right Whales Up-call Detection Using Multimodel Deep Learning

A new method for North Atlantic Right Whales (NARW) up-call detection us...
research
05/21/2021

Distinguishing artefacts: evaluating the saturation point of convolutional neural networks

Prior work has shown Convolutional Neural Networks (CNNs) trained on sur...
research
01/22/2022

Background Invariant Classification on Infrared Imagery by Data Efficient Training and Reducing Bias in CNNs

Even though convolutional neural networks can classify objects in images...
research
06/16/2018

In situ TensorView: In situ Visualization of Convolutional Neural Networks

Convolutional Neural Networks(CNNs) are complex systems. They are traine...

Please sign up or login with your details

Forgot password? Click here to reset