A method to correct binary classifier errors in multi-label pairwise models.
problem Improving the accuracy of binary classifiers in multi-label pairwise models.
method Computing competence and cross-competence measures to estimate and correct label errors.
result The proposed correction methods significantly outperform the reference method in terms of zero-one loss.
Optimal streaming algorithm for crowdsourced binary classification.
problem Binary classification of data using crowdsourcing.
method Learning labeller competence and minimizing prediction error rate.
result Cumulative regret is finite and independent of the number of tasks.
A new weighting scheme corrects label ensembles in multi-label classification.
problem Improving the reliability of multi-label classification with imbalanced data.
method Proposed a novel weighting scheme based on fuzzy confusion matrix and information theory.
result The proposed method reduces the vulnerability to imbalanced class distribution and improves classification quality.
Prototype selection improves DS techniques' accuracy and reduces computational cost.
problem Improving the performance of dynamic selection techniques.
method Prototype selection techniques that edit validation data to remove noise and redundant instances.
result Improves DS techniques' classification accuracy and reduces computational cost.
PrML exploits label relationships using privileged information and low-rank constraints.
problem Improving multi-label learning performance by leveraging implicit and explicit label connections.
method Generates privileged label features and integrates them into low-rank based multi-label learning framework.
result PrML significantly improves multi-label learning performance compared to competing methods.
MCAL reduces labeling costs by 6x for auto-labeling data sets.
problem Expensive human annotation for ground-truth data sets.
method Iterative approach that trains a classifier to auto-label part of the data set, determining which samples to label using humans and which to label using the classifier at each step.
result 6x lower overall cost compared to human labeling the entire data set, always cheaper than competing strategies.
META-DES framework selects best classifiers dynamically.
problem Dynamic selection of ensemble classifiers.
method Meta-learning approach to estimate classifier competence.
result META-DES framework improves classification performance.
Belief Propagation optimally infers true labels from crowdsourced data.
problem Inferring true labels from crowdsourced data with errors.
method Introduced tighter lower bound and proved BP matches it.
result Belief Propagation optimally infers true labels.
Model for detecting rare labels in imbalanced crowdsourcing data.
problem Detecting rare labels in imbalanced crowdsourcing data.
method Generative aggregation model combining item difficulty and class-dependent annotator competence.
result Our model achieves the highest minority recall while maintaining competitive balanced accuracy.
Clarinet uses complementary labels to train classifiers with less source data.
problem Training classifiers with true-label data from source domain is costly.
method Proposes CLARINET to train classifiers with complementary-label source data and unlabeled target data.
result CLARINET significantly outperforms baselines in unsupervised domain adaptation.
CMRM improves robustness in noisy label settings without requiring privileged knowledge.
problem Learning with noisy labels without privileged knowledge.
method Conformal Margin Risk Minimization (CMRM) framework.
result CMRM consistently improves accuracy and reduces mislabeling under various noise conditions.
Neural networks learn patterns in random data, improving downstream performance.
problem Understanding what deep networks learn with random labels.
method Analytical and empirical study of convolutional and fully connected networks pre-trained on random labels.
result Pre-trained networks on random labels transfer faster to real datasets, despite specialization effects.
Paper proposes a method to adapt classifiers using complementary labels instead of true labels.
problem Training classifiers with true labels from the source domain is costly and sometimes impossible.
method Proposes a novel setting with complementary labels and a complementary label adversarial network (CLARINET).
result CLARINET significantly outperforms baselines on handwritten digits and object recognition tasks.
GAML tackles multilabel classification over graphs using message passing and attention.
problem Multilabel classification over graphs with variable-size substructures and label-substructure relations.
method GAML uses a graph neural network that models labels as auxiliary nodes and iteratively applies message passing and attention mechanisms.
result GAML significantly outperforms other methods and provides intuitive visualizations.
HierLPR ranks labels hierarchically for multi-label classification, optimizing a new eAUC metric.
problem Hierarchical multi-label classification with emphasis on first call accuracy.
method Introduces HierLPR algorithm optimizing eAUC metric under tree constraint.
result HierLPR outperforms other methods in early precision-recall curve stages.
Deep pNML improves DNN performance and robustness.
problem Improving deep neural network performance and robustness.
method Introduces pNML scheme for DNNs, extending to twice universal solution.
result pNML outperforms ERM and provides robustness against adversarial attacks.
A method for selecting pseudo-labeled data in semi-supervised learning using generalized Bayes and soft revision.
problem Selecting pseudo-labeled data for semi-supervised learning with robustness to uncertainty.
method Using credal sets and the Gamma-Maximin method with soft revision to update priors and select pseudo-labeled data.
result The Gamma-Maximin method with soft revision can achieve promising results, especially in scenarios with low labeled data proportions.
Deep RL detects anomalies from few labeled examples and large unlabeled data.
problem Anomaly detection with limited labeled data and large unlabeled data.
method Deep reinforcement learning to optimize detection of labeled and unlabeled anomalies.
result Significantly outperforms state-of-the-art methods on 48 real-world datasets.
CrossFilter tackles noisy labels in audio tagging.
problem Noisy labels in large audio datasets.
method CrossFilter framework using multiple representations and multi-task learning.
result Improves audio tagging performance on FSDKaggle2018 and FSDKaggle2019 datasets.
Active WeaSuL uses active learning to improve weak supervision for better model performance.
problem Limited labelled data in machine learning.
method Combines active learning with weak supervision to improve probabilistic labels.
result Active WeaSuL outperforms weak supervision and active learning with limited labelled data.
Unified framework DDNs for multi-label classification, improving inference efficiency.
problem Efficient inference for multi-label classification with dependency networks.
method Combining dependency networks and deep learning, proposing novel inference schemes.
result Novel inference schemes outperform basic neural architectures and Markov networks.
Adversarial robust learning improved for transductive setting.
problem Adversarial robust learning in transductive setting.
method Simple transductive learner for bounded VC dimension classes.
result Robust error rate linear in VC dimension, adaptive to perturbation complexity.
This work explores how to balance rule consistency and coverage for multi-label classification.
problem Balancing rule consistency and coverage for effective multi-label classification.
method Empirical study of rule learning heuristics in multi-label classification.
result The choice of heuristic for multi-label classification is crucial and depends on the performance measure.
Mathematical approach assesses human resource competences accurately.
problem Accurate assessment and representation of human resource competences.
method Detailed quantification scheme and mathematical approach.
result Flexible tools for optimal job assignment and recruitment.
New bounds show unlabeled data can significantly reduce the need for labeled data in semi-supervised multi-objective learning.
problem Achieving good trade-offs in multi-objective learning requires a model with higher capacity, increasing statistical cost.
method Established sample complexity upper bounds using a simple semi-supervised algorithm with pseudo-labeling.
result For some objectives, unlabeled data can significantly alleviate the need for labeled data, reducing statistical cost.
FavMac maximizes value while controlling cost in multi-label prediction.
problem Value-maximizing predictions with strict cost control in multi-label scenarios.
method FavMac pipeline combining any multi-label classifier with online update mechanism.
result FavMac achieves higher value with strict cost control compared to baselines.
Proposes dynamic model type recommendation for OLP technique.
problem Limited local competence of base-classifiers in uneven data distributions.
method Builds a multi-label meta-classifier to recommend model types based on local data complexity.
result Statistically similar performance to original OLP with fixed base-classifier model.
End-to-end anomaly detection framework using labeled anomalies.
problem Limited deep learning for anomaly detection and inefficiency of existing methods.
method Deviation learning neural network with labeled anomalies and prior probability.
result Significantly better anomaly scoring than state-of-the-art methods.
End-to-end framework learns from imperfect annotations directly.
problem Training machine learning models on imperfect human annotations.
method End-to-end framework merging aggregation with model training and modeling annotator competencies.
result Accuracy gains of up to 25% over state-of-the-art annotation aggregation methods.
Amobee's system won 3rd place in Twitter sentiment classification.
problem Sentiment detection on Twitter using deep learning.
method RNN models trained on a sentiment treebank, combined with classifiers.
result 3rd place in SemEval 2017 task 4, 5-label classification.
Paper improves competence estimation of machine learning models.
problem Estimating machine learning performance in real-world scenarios.
method ALICE Score: a pointwise competence estimator considering distributional, data, and model uncertainty.
result Significant improvements in competence prediction over state-of-the-art methods.
Paper evaluates competence measures for DRS systems.
problem Choosing the best measure to quantify competence in DRS systems is challenging.
method Reviewed and adapted eight competence measures for regression problems, compared them on 15 datasets, and evaluated three DRS systems.
result DRS systems outperform individual regressors and static systems, but competence measure choice depends on the problem.
Improved method for encoding contingency tables reduces mutual information bias.
problem Mutual information bias in measuring label similarity.
method Improved method for encoding contingency tables to reduce information cost.
result Better bound on reduced mutual information in typical use cases.
Proposes a new deep ordinal classification model enforcing label uni-modality.
problem Deep ordinal classification with label consistency issues.
method Non-parametric uni-modality constraints via inequality constraints.
result Improves scalability and performance in ordinal classification tasks.
We study the problem of unsupervised domain adaptation, which aims to adapt classifiers trained on a labeled source domain to an unlabeled target domain. Many existing approaches first learn domain-invariant features and then construct classifiers with them. We propose a novel approach that jointly learn the both. Spec…
Proposes KNORA-B and KNORA-BI for DES, improving classification performance.
problem Selecting locally competent classifiers for new test samples.
method KNORA-B and KNORA-BI use nearest neighbors to reduce region of competence, maintaining at least one sample from each class.
result KNORA-BI outperforms state-of-the-art techniques on imbalance datasets.
SSPN uses deep learning to estimate risk scores in survival analysis with competing risks.
problem Nonidentifiability of cause-specific survival curves in competing risk survival analysis.
method Siamese Survival Prognosis Network (SSPN) that avoids estimating cause-specific survival curves and optimizes an approximation to the C-discrimination index.
result SSPN estimates pairwise concordant time-dependent risks, improving risk scoring in survival analysis with competing risks.
This work proposes unsupervised learning by predicting random distances in neural networks.
problem Lack of labelled data in unsupervised learning tasks.
method Train neural networks to predict random distances in a randomly projected space, optimizing for genuine class structures.
result Learned representations outperform state-of-the-art methods in anomaly detection and clustering.
Graph neural networks often assume vertex labels are independent, but we show this is rarely true and propose a method to improve predictions.
problem Graph neural networks often assume vertex labels are conditionally independent given their neighborhood features, which is rarely true.
method We model the joint distribution of residuals on vertices with a parameterized multivariate Gaussian and estimate parameters by maximizing the marginal likelihood of the observed labels.
result Our method achieves substantially higher accuracy than competing baselines and can be interpreted as the strength of correlation among connected vertices.
Comprisk simplifies competing-risks analysis in Python.
problem Analyzing medical time-to-event data with competing risks.
method A scikit-learn-compatible toolkit for competing-risks survival analysis.
result Comprisk provides a unified API for various competing-risks methods.
Improves robust bias-aware prediction under covariate shift.
problem Challenges in machine learning due to distributional shift between source and target data.
method Extends representer theorem to RBA setting, using reweighted kernel expectation.
result Better performance of RBA classifier on synthetic and natural covariate shift datasets.
We developed a framework to benchmark and compare competing risks survival models.
problem Limited systematic evaluation and adoption of competing risks survival models.
method Open-source reproducible benchmarking framework for comprehensive comparison.
result Systematic comparison across multiple datasets on various performance aspects.
New learning scheme outperforms ERM in individual data settings.
problem Learning from individual data samples.
method Information-theoretic approach using self-information loss.
result pNML scheme outperforms ERM in specific test challenges.
SurvivalBoost improves prediction of event times in competing risks scenarios.
problem Predicting event times in scenarios with multiple possible outcomes.
method Developed a strictly proper censoring-adjusted scoring rule for stochastic optimization of competing risks.
result SurvivalBoost outperforms 12 state-of-the-art models across various metrics.
This paper uses neural networks to accurately model competing risks in survival analysis.
problem Ignoring competing risks leads to biased survival estimation in machine learning models.
method The paper introduces constrained monotonic neural networks to model each competing survival distribution.
result The method ensures exact likelihood maximization with reduced computational cost.
Competency questions help experts select best clustering for energy data.
problem Ad hoc and subjective selection of clustering structures by domain experts.
method Formalize expert knowledge and requirements with competency questions.
result Competency questions improve reproducibility and evaluation of clustering applications.
Study examines HTE estimation from time-to-event data with competing events.
problem Estimating HTEs from time-to-event data with competing events.
method Outcome modeling approach using plug-in estimators for potential outcomes.
result Competing events introduce new challenges for HTE estimation.
META-DES.H selects competent classifiers using meta-learning and dynamic weighting.
problem Selecting competent classifiers in dynamic ensemble selection.
method META-DES framework using meta-learning and dynamic weighting.
result Improvements in recognition accuracy on 30 datasets.