This repository contains code to run experiments in the paper "Signal Strength and Noise Drive Feature Preference in CNN Image Classifiers."

Overview

Signal Strength and Noise Drive Feature Preference in CNN Image Classifiers

This repository contains code to run experiments in the paper "Signal Strength and Noise Drive Feature Preference in CNN Image Classifiers." There are three subdirectories in this repository, the contents of which are described below. This code was tested using PyTorch 1.7.

Synthetic Pairs Matrix

This part of the repository is for running the synthetic pairs matrix experiments in the paper. Here are the commands to run all of the experiments in the paper:

Pairs Matrix 1

python main.py --exp_name pairs_matrix1 --pattern_dir pairs_matrix1 --imgnet_augment

Pairs Matrix 2

python main.py --exp_name pairs_matrix2 --pattern_dir pairs_matrix2 --imgnet_augment

Color Deviation

python main.py --exp_name color_deviation_(your epsilon here) --pattern_dir pairs_matrix1 --hue_perturb blue_circle --hue_perturb_val (your epsilon here) --imgnet_augment

Color Overlap (pattern dirs are already predefined for these. Some overlap values are included, but if you would like to use different ones, you must create them yourself.)

python main.py --exp_name color_overlap_(your overlap here) --pattern_dir color_overlap_(your overlap here) --imgnet_augment

Predictivity

python3 main.py --exp_name predictivity_(your predictivity here) --pattern_dir pairs_matrix1 --pred_drop blue --pred_drop_val (your predictivity here)

When you run one of these experiments, datasets will be created and models trained. Datasets will get created and stored in the directory ./data/exp_name, trained models will get stored in ./models/exp_name, and results will appear in ./results/exp_name. When the experiment is done, there should be a file called master.csv in the directory ./results/exp_name which will contain information including each feature's average preference over the course of the experiment, pixel count, and name. A complete list of commands to generate all data in the paper can be found in the commands.sh file in the pairs_matrix_experiments subdirectory. The training script is adapted from the torchvision training script: https://github.com/pytorch/examples/blob/master/imagenet/main.py.

Texture Bias

Stimuli and helper code is used from the open-sourced code of the paper "ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness" (https://github.com/rgeirhos/texture-vs-shape).

To run the experiments from our paper with an ImageNet-trained ResNet-50, you can do the following:

Normal Texture Bias

python main.py

Varying degrees of background interpolation to white (use 0 for completely white, 1 for texture background).

python main.py --bg_interp (your interpolation here)

Resizing

python main.py --bg_interp 0 --size (your fraction of the object size here)

Landscapes

python main.py --bg_interp 0 --landscape

Only full shapes

python main.py --only_complete

Only full shapes masked with masked/interpolated background

python main.py --only_complete --bg_interp (your interpolation here)

A complete list of commands to generate all of the texture bias data from our paper can be found in the commands.sh file in the texture_bias subdirectory.

Excessive Invariance

Running these experiments is a bit more involved. A complete list of commands you must run to reproduce all data and graphs found in the paper can be found in the commands.sh file in the excessive_invariance subdirectory. Comments in the file describe what each step represents.

PyTorch implementation of "VRT: A Video Restoration Transformer"

VRT: A Video Restoration Transformer Jingyun Liang, Jiezhang Cao, Yuchen Fan, Kai Zhang, Rakesh Ranjan, Yawei Li, Radu Timofte, Luc Van Gool Computer

Jingyun Liang 837 Jan 09, 2023
Foreground-Action Consistency Network for Weakly Supervised Temporal Action Localization

FAC-Net Foreground-Action Consistency Network for Weakly Supervised Temporal Action Localization Linjiang Huang (CUHK), Liang Wang (CASIA), Hongsheng

21 Nov 22, 2022
Code for the paper "There is no Double-Descent in Random Forests"

Code for the paper "There is no Double-Descent in Random Forests" This repository contains the code to run the experiments for our paper called "There

2 Jan 14, 2022
Permeability Prediction Via Multi Scale 3D CNN

Permeability-Prediction-Via-Multi-Scale-3D-CNN Data: The raw CT rock cores are obtained from the Imperial Colloge portal. The CT rock cores are sub-sa

Mohamed Elmorsy 2 Jul 06, 2022
Free-duolingo-plus - Duolingo account creator that uses your invite code to get you free duolingo plus

free-duolingo-plus duolingo account creator that uses your invite code to get yo

1 Jan 06, 2022
A transformer-based method for Healthcare Image Captioning in Vietnamese

vieCap4H Challenge 2021: A transformer-based method for Healthcare Image Captioning in Vietnamese This repo GitHub contains our solution for vieCap4H

Doanh B C 4 May 05, 2022
BanditPAM: Almost Linear-Time k-Medoids Clustering

BanditPAM: Almost Linear-Time k-Medoids Clustering This repo contains a high-performance implementation of BanditPAM from BanditPAM: Almost Linear-Tim

254 Dec 12, 2022
A Broader Picture of Random-walk Based Graph Embedding

Random-walk Embedding Framework This repository is a reference implementation of the random-walk embedding framework as described in the paper: A Broa

Zexi Huang 23 Dec 13, 2022
gACSON software for visualization, processing and analysis of three-dimensional electron microscopy images

gACSON gACSON software is to visualize, segment, and analyze the morphology of neurons in three-dimensional electron microscopy images. If you use any

Andrea Behanova 2 May 31, 2022
A python toolbox for predictive uncertainty quantification, calibration, metrics, and visualization

Website, Tutorials, and Docs    Uncertainty Toolbox A python toolbox for predictive uncertainty quantification, calibration, metrics, and visualizatio

Uncertainty Toolbox 1.4k Dec 28, 2022
Repository for the electrical and ICT benchmark model developed in the ERIGrid 2.0 project.

Benchmark Model Electrical and ICT System This repository contains the documentation, code, and models for the electrical and ICT benchmark model deve

ERIGrid 2.0 1 Nov 29, 2021
Baselines for TrajNet++

TrajNet++ : The Trajectory Forecasting Framework PyTorch implementation of Human Trajectory Forecasting in Crowds: A Deep Learning Perspective TrajNet

VITA lab at EPFL 183 Jan 05, 2023
Exploring Versatile Prior for Human Motion via Motion Frequency Guidance (3DV2021)

Exploring Versatile Prior for Human Motion via Motion Frequency Guidance This is the codebase for video-based human motion reconstruction in human-mot

Jiachen Xu 5 Jul 14, 2022
A tensorflow/keras implementation of StyleGAN to generate images of new Pokemon.

PokeGAN A tensorflow/keras implementation of StyleGAN to generate images of new Pokemon. Dataset The model has been trained on dataset that includes 8

19 Jul 26, 2022
Run Keras models in the browser, with GPU support using WebGL

**This project is no longer active. Please check out TensorFlow.js.** The Keras.js demos still work but is no longer updated. Run Keras models in the

Leon Chen 4.9k Dec 29, 2022
MIM: MIM Installs OpenMMLab Packages

MIM provides a unified API for launching and installing OpenMMLab projects and their extensions, and managing the OpenMMLab model zoo.

OpenMMLab 254 Jan 04, 2023
Compute execution plan: A DAG representation of work that you want to get done. Individual nodes of the DAG could be simple python or shell tasks or complex deeply nested parallel branches or embedded DAGs themselves.

Hello from magnus Magnus provides four capabilities for data teams: Compute execution plan: A DAG representation of work that you want to get done. In

12 Feb 08, 2022
Code to train models from "Paraphrastic Representations at Scale".

Paraphrastic Representations at Scale Code to train models from "Paraphrastic Representations at Scale". The code is written in Python 3.7 and require

John Wieting 71 Dec 19, 2022
FastFace: Lightweight Face Detection Framework

Light Face Detection using PyTorch Lightning

Ömer BORHAN 75 Dec 05, 2022
To Design and Implement Logistic Regression to Classify Between Benign and Malignant Cancer Types

To Design and Implement Logistic Regression to Classify Between Benign and Malignant Cancer Types, from a Database Taken From Dr. Wolberg reports his Clinic Cases.

Astitva Veer Garg 1 Jul 31, 2022