Keep CALM and Improve Visual Feature Attribution

NAVER AI

Last update: Dec 7, 2022

Related tags

Deep Learning calm

Overview

Keep CALM and Improve Visual Feature Attribution

Jae Myung Kim^1*, Junsuk Choe^1*, Zeynep Akata², Seong Joon Oh^1†
_{* Equal contribution} _† _{Corresponding author}

¹ _{NAVER AI LAB} ² _{University of Tübingen}

Abstract

The class activation mapping, or CAM, has been the cornerstone of feature attribution methods for multiple vision tasks. Its simplicity and effectiveness have led to wide applications in the explanation of visual predictions and weakly-supervised localization tasks. However, CAM has its own shortcomings. The computation of attribution maps relies on ad-hoc calibration steps that are not part of the training computational graph, making it difficult for us to understand the real meaning of the attribution values. In this paper, we improve CAM by explicitly incorporating a latent variable encoding the location of the cue for recognition in the formulation, thereby subsuming the attribution map into the training computational graph. The resulting model, class activation latent mapping, or CALM, is trained with the expectation-maximization algorithm. Our experiments show that CALM identifies discriminative attributes for image classifiers more accurately than CAM and other visual attribution baselines. CALM also shows performance improvements over prior arts on the weakly-supervised object localization benchmarks.

Dataset downloading

For ImageNet and CUB datasets, please follow the common procedure for downloading the datasets.
For ImageNetV2, CUBV2, and OpenImages30k, please follow the procedure introduced in wsol-evaluation page.

How to use models

You can train CALM models by

$ python main.py --experiment_name=experiment_name/ \
                 --architecture=resnet50 \
                 --attribution_method=CALM_EM \
                 --dataset=CUB \
                 --use_bn=True --large_feature_map=True

You can evaluate the models on two different metrics,

$ python eval_pixel_perturb.py --experiment_name=experiment_name/ \
                               --architecture=resnet50 \ 
                               --attribution_method=CALM_EM \
                               --dataset=CUB \
                               --use_bn=True --large_feature_map=True \
                               --use_load_checkpoint=True \
                               --load_checkpoint=checkpoint_name/ \
                               --score_map_process=jointll --norm_type=clipping &
                               
$ python eval_cue_location.py --experiment_name=experiment_name/ \ 
                              --architecture=resnet50 \
                              --attribution_method=CALM_EM \
                              --dataset=CUB \
                              --use_bn=True --large_feature_map=True \
                              --use_load_checkpoint=True \
                              --load_checkpoint=checkpoint_name/ \
                              --score_map_process=jointll --norm_type=clipping --threshold_type=log &

Pretrained weights

For those who wish to use pretrained CALM weights,

Model name	Dataset	cls. accuracy	weights
CALM_EM	CUB	71.8	link
CALM_EM	OpenImages	70.1	link
CALM_EM	ImageNet	70.4	link
CALM_ML	CUB	59.6	link
CALM_ML	OpenImages	70.9	link
CALM_ML	ImageNet	70.6	link

Explainability scores

Cue localization and Remove-and-classify results. More details about the metrics are in the paper.

Cue localization (the higher, the better)	Remove-and-classify (the lower, the better)

License

Copyright (c) 2021-present NAVER Corp.

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.

Comments

How to calculate the loss in the EM setting?

Hi, Thank you for your great work, it is really interesting and complicated too. I still do not understand how you calculate the loss in the EM setting, you said in the paper that this is a self-supervised setting in the EM case, and calculates the pixel-wise nll between the pseudo-target and the joint likelihood block. But I found in your code, you still use the image label to input to the nll loss, and here did you take the sum over two last channels which will output a specific vector and use it to compare which the image label? If this is what you have done, it will contradict what you said in the paper, right?

The second one is the attribution map of p(z,y^|x) and p(y^|x,z) here are similar, so this violate the equation p(y, z|x) = p(y|x,z)p(z|x), right? Can you also clarify on this, why these two maps are so similar?

Hope you can clarify these, they are the things I am thinking about a lot.

Best, Tin

opened by ngthanhtin 7

improvement of CLIP features over the traditional resnet features on the visual question answering, image captioning, navigation and visual entailment tasks.

CLIP-ViL In our paper "How Much Can CLIP Benefit Vision-and-Language Tasks?", we show the improvement of CLIP features over the traditional resnet fea

310 Dec 28, 2022

Official codebase for Legged Robots that Keep on Learning: Fine-Tuning Locomotion Policies in the Real World

Legged Robots that Keep on Learning Official codebase for Legged Robots that Keep on Learning: Fine-Tuning Locomotion Policies in the Real World, whic

70 Dec 7, 2022

Monify: an Expense tracker Program implemented in a Graphical User Interface that allows users to keep track of their expenses

💳 MONIFY (EXPENSE TRACKER PRO) 💳 Description Monify is an Expense tracker Program implemented in a Graphical User Interface allows users to add inco

1 Dec 14, 2021

Use graph-based analysis to re-classify stocks and to improve Markowitz portfolio optimization

Dynamic Stock Industrial Classification Use graph-based analysis to re-classify stocks and experiment different re-classification methodologies to imp

10 Dec 5, 2022

Deep Neural Networks Improve Radiologists' Performance in Breast Cancer Screening

Deep Neural Networks Improve Radiologists' Performance in Breast Cancer Screening Introduction This is an implementation of the model used for breast

757 Dec 30, 2022

Official implementation of "Open-set Label Noise Can Improve Robustness Against Inherent Label Noise" (NeurIPS 2021)

Open-set Label Noise Can Improve Robustness Against Inherent Label Noise NeurIPS 2021: This repository is the official implementation of ODNL. Require

12 Dec 7, 2022

SmallInitEmb - LayerNorm(SmallInit(Embedding)) in a Transformer to improve convergence

SmallInitEmb LayerNorm(SmallInit(Embedding)) in a Transformer I find that when t

11 Dec 25, 2022

Models Supported: AlbUNet [18, 34, 50, 101, 152] (1D and 2D versions for Single and Multiclass Segmentation, Feature Extraction with supports for Deep Supervision and Guided Attention)

AlbUNet-1D-2D-Tensorflow-Keras This repository contains 1D and 2D Signal Segmentation Model Builder for AlbUNet and several of its variants developed

1 Nov 15, 2021

Episodic Transformer (E.T.) is a novel attention-based architecture for vision-and-language navigation. E.T. is based on a multimodal transformer that encodes language inputs and the full episode history of visual observations and actions.

Episodic Transformers (E.T.) Episodic Transformer for Vision-and-Language Navigation Alexander Pashevich, Cordelia Schmid, Chen Sun Episodic Transform

62 Dec 24, 2022

Keep CALM and Improve Visual Feature Attribution

Related tags

Overview

Keep CALM and Improve Visual Feature Attribution

Abstract

Dataset downloading

How to use models

Pretrained weights

Explainability scores

License

You might also like...

improvement of CLIP features over the traditional resnet features on the visual question answering, image captioning, navigation and visual entailment tasks.

Official codebase for Legged Robots that Keep on Learning: Fine-Tuning Locomotion Policies in the Real World

Monify: an Expense tracker Program implemented in a Graphical User Interface that allows users to keep track of their expenses

Use graph-based analysis to re-classify stocks and to improve Markowitz portfolio optimization

Deep Neural Networks Improve Radiologists' Performance in Breast Cancer Screening

Official implementation of "Open-set Label Noise Can Improve Robustness Against Inherent Label Noise" (NeurIPS 2021)

SmallInitEmb - LayerNorm(SmallInit(Embedding)) in a Transformer to improve convergence

Models Supported: AlbUNet [18, 34, 50, 101, 152] (1D and 2D versions for Single and Multiclass Segmentation, Feature Extraction with supports for Deep Supervision and Guided Attention)

Episodic Transformer (E.T.) is a novel attention-based architecture for vision-and-language navigation. E.T. is based on a multimodal transformer that encodes language inputs and the full episode history of visual observations and actions.

Comments

How to calculate the loss in the EM setting?

Owner

NAVER AI

PyTorch implementation of VAGAN: Visual Feature Attribution Using Wasserstein GANs

An open source app to help calm you down when needed.

The official repository for our paper "The Devil is in the Detail: Simple Tricks Improve Systematic Generalization of Transformers". We significantly improve the systematic generalization of transformer models on a variety of datasets using simple tricks and careful considerations.

git《Self-Attention Attribution: Interpreting Information Interactions Inside Transformer》(AAAI 2021) GitHub:

This is the pytorch implementation of the paper - Axiomatic Attribution for Deep Networks.

This is the pytorch implementation for the paper: Generalizable Mixed-Precision Quantization via Attribution Rank Preservation, which is accepted to ICCV2021.

Text mining project; Using distilBERT to predict authors in the classification task authorship attribution.

pytorch implementation of "Contrastive Multiview Coding", "Momentum Contrast for Unsupervised Visual Representation Learning", and "Unsupervised Feature Learning via Non-Parametric Instance-level Discrimination"

NFT-Price-Prediction-CNN - Using visual feature extraction, prices of NFTs are predicted via CNN (Alexnet and Resnet) architectures.

naked is a Python tool which allows you to strip a model and only keep what matters for making predictions.