Deep Bayesian Active Learning with Image Data

Abstract

Even though active learning forms an important pillar of machine learning, deep learning tools are not prevalent within it. Deep learning poses several difficulties when used in an active learning setting. First, active learning (AL) methods generally rely on being able to learn and update models from small amounts of data. Recent advances in deep learning, on the other hand, are notorious for their dependence on large amounts of data. Second, many AL acquisition functions rely on model uncertainty, yet deep learning methods rarely represent such model uncertainty. In this paper we combine recent advances in Bayesian deep learning into the active learning framework in a practical way. We develop an active learning framework for high dimensional data, a task which has been extremely challenging so far, with very sparse existing literature. Taking advantage of specialised models such as Bayesian convolutional neural networks, we demonstrate our active learning techniques with image data, obtaining a significant improvement on existing active learning approaches. We demonstrate this on both the MNIST dataset, as well as for skin cancer diagnosis from lesion images (ISIC2016 task).

References in corpus (7)

• Sequence to Sequence Learning with Neural Networks
• Improving neural networks by preventing co-adaptation of feature detectors
• Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning
• Exploring the Limits of Language Modeling
• Bayesian Convolutional Neural Networks with Bernoulli Approximate Variational Inference
• Bayesian Active Learning for Classification and Preference Learning
• Distributional Smoothing with Virtual Adversarial Training

Cited by in corpus (113)

• A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challenges
• Hands-on Bayesian Neural Networks -- a Tutorial for Deep Learning Users
• Active Learning for Convolutional Neural Networks: A Core-Set Approach
• A Survey on Active Learning and Human-in-the-Loop Deep Learning for Medical Image Analysis
• A Survey of Deep RL and IL for Autonomous Driving Policy Learning
• Adversarial Active Learning for Deep Networks: a Margin Based Approach
• Bayesian semi-supervised learning for uncertainty-calibrated prediction of molecular properties and active learning
• Half a Percent of Labels is Enough: Efficient Animal Detection in UAV Imagery using Deep CNNs and Active Learning
• Diminishing Uncertainty within the Training Pool: Active Learning for Medical Image Segmentation
• Discriminative Active Learning
• Active label cleaning for improved dataset quality under resource constraints
• Deep Active Learning over the Long Tail
• Active, Continual Fine Tuning of Convolutional Neural Networks for Reducing Annotation Efforts
• Bayesian Batch Active Learning as Sparse Subset Approximation
• TensorLayer: A Versatile Library for Efficient Deep Learning Development
• QU-BraTS: MICCAI BraTS 2020 Challenge on Quantifying Uncertainty in Brain Tumor Segmentation - Analysis of Ranking Scores and Benchmarking Results
• Computer Vision with Deep Learning for Plant Phenotyping in Agriculture: A Survey
• Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safety
• Model Selection in Bayesian Neural Networks via Horseshoe Priors
• Out-of-distribution Detection in Classifiers via Generation
• Dropout-based Active Learning for Regression
• Active Learning with Partial Feedback
• Mapping oil palm density at country scale: An active learning approach
• Deep Active Learning for Named Entity Recognition
• Transient-optimised real-bogus classification with Bayesian Convolutional Neural Networks -- sifting the GOTO candidate stream
• O-MedAL: Online Active Deep Learning for Medical Image Analysis
• Embedded deep learning in ophthalmology: Making ophthalmic imaging smarter
• Detecting Out-of-Distribution Inputs in Deep Neural Networks Using an Early-Layer Output
• Active Learning for Segmentation by Optimizing Content Information for Maximal Entropy
• Adversarial Distillation of Bayesian Neural Network Posteriors
• SPINBIS: Spintronics based Bayesian Inference System with Stochastic Computing
• Large-Scale Visual Active Learning with Deep Probabilistic Ensembles
• Efficient Personalized Speech Enhancement through Self-Supervised Learning
• Deep Indexed Active Learning for Matching Heterogeneous Entity Representations
• Deep Active Learning for Computer Vision: Past and Future
• Structured Variational Learning of Bayesian Neural Networks with Horseshoe Priors
• Deep Ensemble Bayesian Active Learning : Addressing the Mode Collapse issue in Monte Carlo dropout via Ensembles
• CEREALS - Cost-Effective REgion-based Active Learning for Semantic Segmentation
• A semi-supervised self-training method to develop assistive intelligence for segmenting multiclass bridge elements from inspection videos
• Reinforced active learning for image segmentation
• The Relevance of Bayesian Layer Positioning to Model Uncertainty in Deep Bayesian Active Learning
• Towards Active Learning for Action Spotting in Association Football Videos
• Active and Incremental Learning with Weak Supervision
• Interpolating Detailed Simulations of Kilonovae: Adaptive Learning and Parameter Inference Applications
• Using Error Decay Prediction to Overcome Practical Issues of Deep Active Learning for Named Entity Recognition
• Wat zei je? Detecting Out-of-Distribution Translations with Variational Transformers
• Model-Centric and Data-Centric Aspects of Active Learning for Deep Neural Networks
• Active Learning for Visual Question Answering: An Empirical Study
• Deeper Connections between Neural Networks and Gaussian Processes Speed-up Active Learning
• Bayesian Active Meta-Learning for Few Pilot Demodulation and Equalization
• Self-supervised Assisted Active Learning for Skin Lesion Segmentation
• MedAL: Deep Active Learning Sampling Method for Medical Image Analysis
• Deep Active Learning for Axon-Myelin Segmentation on Histology Data
• Minimax Active Learning
• Bayesian Semisupervised Learning with Deep Generative Models
• Neural Simpletrons - Minimalistic Directed Generative Networks for Learning with Few Labels
• A Mathematical Analysis of Learning Loss for Active Learning in Regression
• Active Learning to Classify Macromolecular Structures in situ for Less Supervision in Cryo-Electron Tomography
• Minority Class Oriented Active Learning for Imbalanced Datasets
• Deep Active Learning for Text Classification with Diverse Interpretations
• Large Margin Deep Networks for Classification
• Exploring Spatial Diversity for Region-based Active Learning
• Active Learning of Abstract Plan Feasibility
• Active Image Synthesis for Efficient Labeling
• Treatment Learning Causal Transformer for Noisy Image Classification
• Boosting Semi-Supervised Object Detection in Remote Sensing Images With Active Teaching
• Temporal Output Discrepancy for Loss Estimation-based Active Learning
• Bayesian Active Learning for Discrete Latent Variable Models
• Uncertainty in Multitask Transfer Learning
• Novel Batch Active Learning Approach and Its Application to Synthetic Aperture Radar Datasets
• VAE-Info-cGAN: Generating Synthetic Images by Combining Pixel-level and Feature-level Geospatial Conditional Inputs
• Deep Bayesian Active Learning, A Brief Survey on Recent Advances
• Active learning using weakly supervised signals for quality inspection
• Model-Change Active Learning in Graph-Based Semi-Supervised Learning
• Deep Active Learning for Efficient Training of a LiDAR 3D Object Detector
• On the use of uncertainty in classifying Aedes Albopictus mosquitoes
• VFunc: a Deep Generative Model for Functions
• Iterate & Cluster: Iterative Semi-Supervised Action Recognition
• Unsupervised Data Uncertainty Learning in Visual Retrieval Systems
• Active learning for object detection in high-resolution satellite images
• Towards Balanced Active Learning for Multimodal Classification
• Collaborative Intelligence Orchestration: Inconsistency-Based Fusion of Semi-Supervised Learning and Active Learning
• Quantification of Uncertainties in Deep Learning-based Environment Perception
• Message Passing Adaptive Resonance Theory for Online Active Semi-supervised Learning
• Improved Synthetic Training for Reading Comprehension
• Cost-Effective Training of Deep CNNs with Active Model Adaptation
• Are Fewer Labels Possible for Few-shot Learning?
• On the Robustness of Active Learning
• Iterative Projection and Matching: Finding Structure-preserving Representatives and Its Application to Computer Vision
• Importance of Self-Consistency in Active Learning for Semantic Segmentation
• One-bit Supervision for Image Classification: Problem, Solution, and Beyond
• Ask-n-Learn: Active Learning via Reliable Gradient Representations for Image Classification
• Pool-Based Unsupervised Active Learning for Regression Using Iterative Representativeness-Diversity Maximization (iRDM)
• Toward Machine-Guided, Human-Initiated Explanatory Interactive Learning
• Query-based Hard-Image Retrieval for Object Detection at Test Time
• Calpric: Inclusive and Fine-grain Labeling of Privacy Policies with Crowdsourcing and Active Learning
• Bayesian Uncertainty and Expected Gradient Length -- Regression: Two Sides Of The Same Coin?
• Looking at the posterior: accuracy and uncertainty of neural-network predictions
• Robust Assignment of Labels for Active Learning with Sparse and Noisy Annotations
• Analyzing Data Selection Techniques with Tools from the Theory of Information Losses
• Alpha-Divergences in Variational Dropout
• A Statistician Teaches Deep Learning
• Episode-Based Active Learning with Bayesian Neural Networks
• Leveraging Local Variation in Data: Sampling and Weighting Schemes for Supervised Deep Learning
• Bias-Aware Heapified Policy for Active Learning
• Assisted Robust Reward Design
• Stochastic Bayesian Neural Networks
• Improve Unsupervised Pretraining for Few-label Transfer
• Uncertainty Estimation in Deep Neural Networks for Point Cloud Segmentation in Factory Planning
• Leveraging Motion Priors in Videos for Improving Human Segmentation
• The Effect of Optimization Methods on the Robustness of Out-of-Distribution Detection Approaches
• Information Planning for Text Data
• Uncertainty-Aware Data Aggregation for Deep Imitation Learning