Generalized Parametric Contrastive Learning (2022-09-26T00:00:00.000000Z)

TL;DR

This analysis demonstrates that GPaCo/PaCo can adaptively enhance the intensity of pushing samples of the same class close as more samples are pulled together with their corresponding centers and benefit hard example learning.

Abstract

In this paper, we propose the Generalized Parametric Contrastive Learning (GPaCo/PaCo) which works well on both imbalanced and balanced data. Based on theoretical analysis, we observe supervised contrastive loss tends to bias on high-frequency classes and thus increases the difficulty of imbalanced learning. We introduce a set of parametric class-wise learnable centers to rebalance from an optimization perspective. Further, we analyze our GPaCo/PaCo loss under a balanced setting. Our analysis demonstrates that GPaCo/PaCo can adaptively enhance the intensity of pushing samples of the same class close as more samples are pulled together with their corresponding centers and benefit hard example learning. Experiments on long-tailed benchmarks manifest the new state-of-the-art for long-tailed recognition. On full ImageNet, models from CNNs to vision transformers trained with GPaCo loss show better generalization performance and stronger robustness compared with MAE models. Moreover, GPaCo can be applied to semantic segmentation task and obvious improvements are observed on 4 most popular benchmarks.

Authors

Jiaya Jia

23 papers

Bei Yu

3 papers

Zhuotao Tian

4 papers

TL;DR

Abstract

Authors

References71 items

Region Rebalance for Long-Tailed Semantic Segmentation

Masked Autoencoders Are Scalable Vision Learners

Contrastive Learning With Stronger Augmentations

ResLT: Residual Learning for Long-Tailed Recognition

Exploring Simple Siamese Representation Learning

An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

Long-tailed Recognition by Routing Diverse Distribution-Aware Experts

Improving Calibration for Long-Tailed Recognition

Long-Tailed Classification by Keeping the Good and Removing the Bad Momentum Causal Effect

Balanced Meta-Softmax for Long-Tailed Visual Recognition

The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalization

ELF: An Early-Exiting Framework for Long-Tailed Classification

Unsupervised Learning of Visual Features by Contrasting Cluster Assignments

Bootstrap Your Own Latent: A New Approach to Self-Supervised Learning

Supervised Contrastive Learning

Rethinking Class-Balanced Methods for Long-Tailed Visual Recognition From a Domain Adaptation Perspective

Equalization Loss for Long-Tailed Object Recognition

Improved Baselines with Momentum Contrastive Learning

A Simple Framework for Contrastive Learning of Visual Representations

BBN: Bilateral-Branch Network With Cumulative Learning for Long-Tailed Visual Recognition

Momentum Contrast for Unsupervised Visual Representation Learning

Decoupling Representation and Classifier for Long-Tailed Recognition

Fast and Practical Neural Architecture Search

Randaugment: Practical automated data augmentation with a reduced search space

Once for All: Train One Network and Specialize it for Efficient Deployment

Learning Imbalanced Datasets with Label-Distribution-Aware Margin Loss

AutoAugment: Learning Augmentation Strategies From Data

Learning Robust Global Representations by Penalizing Local Predictive Power

Large-Scale Long-Tailed Recognition in an Open World

Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting

Class-Balanced Loss Based on Effective Number of Samples

What is the Effect of Importance Weighting in Deep Learning?

MnasNet: Platform-Aware Neural Architecture Search for Mobile

ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design

Unified Perceptual Parsing for Scene Understanding

Representation Learning with Contrastive Predictive Coding

Benchmarking Neural Network Robustness to Common Corruptions and Surface Variations

DARTS: Differentiable Architecture Search

Deep Imbalanced Learning for Face Recognition and Attribute Prediction

Places: A 10 Million Image Database for Scene Recognition

Piecewise Classifier Mappings: Learning Fine-Grained Learners for Novel Categories With Few Examples

Unsupervised Feature Learning via Non-parametric Instance Discrimination

Learning to Reweight Examples for Robust Deep Learning

Path Aggregation Network for Instance Segmentation

A systematic study of the class imbalance problem in convolutional neural networks

Improved Regularization of Convolutional Neural Networks with Cutout

Learning Transferable Architectures for Scalable Image Recognition

Scene Parsing through ADE20K Dataset

The iNaturalist Species Classification and Detection Dataset

COCO-Stuff: Thing and Stuff Classes in Context

Feature Pyramid Networks for Object Detection

Pyramid Scene Parsing Network

Learning Deep Representation for Imbalanced Classification

The Cityscapes Dataset for Semantic Urban Scene Understanding

Factors in Finetuning Deep Model for Object Detection with Long-Tail Distribution

Relay Backpropagation for Effective Learning of Deep Convolutional Neural Networks

Deep Residual Learning for Image Recognition

Very Deep Convolutional Networks for Large-Scale Image Recognition

ImageNet Large Scale Visual Recognition Challenge

The Role of Context for Object Detection and Semantic Segmentation in the Wild

Microsoft COCO: Common Objects in Context

The class imbalance problem: A systematic study

Swin Transformer: Hierarchical Vision Transformer using Shifted Windows

“MMSegmentation: OpenMMLab semantic segmentation toolbox and benchmark,”

Learning From Imbalanced Data

Learning to Model the Tail

“Towards good practices for recognition & detection,”

Learning Multiple Layers of Features from Tiny Images

SMOTE: Synthetic Minority Over-sampling Technique

,in2022,underthesupervisionofProf.

is with the SmartMore, Shenzhen 154100, China

Field of Study

Journal Information

Name

Page

Volume