Multi-Label Learning from Single Positive Labels (2021-06-01T00:00:00.000000Z)

TL;DR

This work considers the hardest version of this problem, where annotators provide only one relevant label for each image, and shows that in some cases it is possible to approach the performance of fully labeled classifiers despite training with significantly fewer confirmed labels.

Abstract

Predicting all applicable labels for a given image is known as multi-label classification. Compared to the standard multi-class case (where each image has only one label), it is considerably more challenging to annotate training data for multi-label classification. When the number of potential labels is large, human annotators find it difficult to mention all applicable labels for each training image. Furthermore, in some settings detection is intrinsically difficult e.g. finding small object instances in high resolution images. As a result, multi-label training data is often plagued by false negatives. We consider the hardest version of this problem, where annotators provide only one relevant label for each image. As a result, training sets will have only one positive label per image and no confirmed negatives. We explore this special case of learning from missing labels across four different multi-label image classification datasets for both linear classifiers and end-to-end fine-tuned deep networks. We extend existing multi-label losses to this setting and propose novel variants that constrain the number of expected positive labels during training. Surprisingly, we show that in some cases it is possible to approach the performance of fully labeled classifiers despite training with significantly fewer confirmed labels.

Authors

P. Perona

11 papers

Oisin Mac Aodha

6 papers

Dan Morris

2 papers

TL;DR

Abstract

Authors

References68 items

Set-valued classification - overview via a unified framework

The Emerging Trends of Multi-Label Learning

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

Interactive Multi-Label CNN Learning With Partial Labels

Does label smoothing mitigate label noise?

Survey on Multi-Output Learning

Multi-label Co-regularization for Semi-supervised Facial Action Unit Recognition

Presence-Only Geographical Priors for Fine-Grained Image Classification

When Does Label Smoothing Help?

LVIS: A Dataset for Large Vocabulary Instance Segmentation

Weakly Supervised Image Classification Through Noise Regularization

Elucidating image-to-set prediction: An analysis of models, losses and datasets

Multi-Label Image Recognition With Graph Convolutional Networks

Learning Multi Labels from Single Label —— An Extreme Weak Label Learning Algorithm

Learning a Deep ConvNet for Multi-Label Classification With Partial Labels

Tencent ML-Images: A Large-Scale Multi-Label Image Database for Visual Representation Learning

Multi-label Learning from Noisy Labels with Non-linear Feature Transformation

Learning from positive and unlabeled data: a survey

Deep Generative Models for Weakly-Supervised Multi-Label Classification

Multi-label Learning with Highly Incomplete Data via Collaborative Embedding

Exploring the Limits of Weakly Supervised Pretraining

Advances in deep learning approaches for image tagging

Deep Learning for Extreme Multi-label Text Classification

The iNaturalist Species Classification and Detection Dataset

Deep Learning is Robust to Massive Label Noise

Learning from Complementary Labels

Nonconvex One-bit Single-label Multi-label Learning

Learning from Noisy Large-Scale Datasets with Minimal Supervision

Understanding deep learning requires rethinking generalization

Bag of Tricks for Efficient Text Classification

Multi-label Ranking from Positive and Unlabeled Data

CNN-RNN: A Unified Framework for Multi-label Image Classification

Deep Residual Learning for Image Recognition

Rethinking the Inception Architecture for Computer Vision

Learning Visual Features from Large Weakly Supervised Data

Polyphonic sound event detection using multi label deep neural networks

Effective Use of Word Order for Text Categorization with Convolutional Neural Networks

PU Learning for Matrix Completion

ImageNet Large Scale Visual Recognition Challenge

FastXML: a fast, accurate and stable tree-classifier for extreme multi-label learning

A Review on Multi-Label Learning Algorithms

HCP: A Flexible CNN Framework for Multi-Label Image Classification

Microsoft COCO: Common Objects in Context

Scalable multi-label annotation

Large-Scale Multi-label Text Classification - Revisiting Neural Networks

Deep Convolutional Ranking for Multilabel Image Annotation

Speedup Matrix Completion with Side Information: Application to Multi-Label Learning

Dynamic Label Propagation for Semi-supervised Multi-class Multi-label Classification

Multilabel Classification using Bayesian Compressed Sensing

Acoustic classification of multiple simultaneous bird species: a multi-instance multi-label approach.

User-Aware Image Tag Refinement via Ternary Semantic Analysis

Matrix Completion for Multi-label Image Classification

The Caltech-UCSD Birds-200-2011 Dataset

Multi-label learning with incomplete class assignments

Multi-Label Learning with Weak Label

NUS-WIDE: a real-world web image database from National University of Singapore

Multi-Label Prediction via Compressed Sensing

ML-KNN: A lazy learning approach to multi-label learning

Semi-supervised Multi-label Learning by Constrained Non-negative Matrix Factorization

Cognitive psychology: Rare items often missed in visual searches

A maximum entropy approach to species distribution modeling

Learning to Classify Texts Using Positive and Unlabeled Data

Suppl. Material: Exploiting weakly supervised visual patterns to learn from partial annotations

The iNaturalist Species Classification and Detection Dataset-Supplementary Material

Visual search

The PASCAL Visual Object Classes Challenge

The Pascal Visual Object Classes Challenge 2006 ( VOC 2006 ) Results

Pascal Visual Object Classes Challenge Results

Field of Study

Journal Information

Name

Page

Venue Information

Name

Type

URL