Distillation improves simple models by approximating complex labels.
problem Why does distillation improve simple models?
method Statistical perspective on distillation, connecting to extreme multiclass retrieval.
result Distillation helps by approximating underlying class-probabilities, reducing bias and variance.
The law of total probability may be deployed in binary classification exercises to estimate the unconditional class probabilities if the class proportions in the training set are not representative of the population class proportions. We argue that this is not a conceptually sound approach and suggest an alternative ba…
Knowledge distillation improves model accuracy by mimicking teacher model probabilities.
problem Improving model accuracy through model compression.
method Casting knowledge distillation as a semiparametric inference problem, deriving new guarantees, and developing enhancements.
result Enhancements improve student performance by mitigating teacher overfitting and underfitting.
A new method for adapting to label shifts using class probability matching.
problem Adapting to label shifts where class probabilities differ between source and target domains.
method Class Probability Matching using Kernel Methods (CPMKM) framework.
result CPMKM outperforms existing methods on real datasets.
This paper proposes a methodology for host-based anomaly detection using a semi-supervised algorithm namely one-class classifier combined with a PCA-based feature extraction technique called Eigentraces on system call trace data. The one-class classification is based on generating a set of artificial data using a refer…
In this work we investigate to which extent one can recover class probabilities within the empirical risk minimization (ERM) paradigm. The main aim of our paper is to extend existing results and emphasize the tight relations between empirical risk minimization and class probability estimation. Based on existing literat…
Study post-hoc Learning to Defer using density-ratio losses.
problem Optimizing decision-making between models and experts.
method Density-ratio losses for post-hoc L2D scorers, derived from class-probability estimation.
result The approach recovers known results and introduces new connections to expert comparison and anomaly detection.
We study the problem of supervised learning for both binary and multiclass classification from a unified geometric perspective. In particular, we propose a geometric regularization technique to find the submanifold corresponding to a robust estimator of the class probability P(y∣x). The regularization term meas…
An imprecise SHAP method explains class probabilities with limited data.
problem Explaining class probabilities with limited training data.
method New approach for computing feature marginal contributions and general approach to interval-valued Shapley values.
result The imprecise SHAP method improves explanation of class probabilities.
Paper introduces a novel method for estimating model confidence in deep neural classifiers.
problem Reliable confidence estimation for deep neural classifiers in safety-critical applications.
method Proposes a novel target criterion (true class probability) and learns it from data with an auxiliary model.
result The proposed method outperforms strong baselines in various tasks and network architectures.
A new method detects changes in multivariate data using random forests.
problem Detecting changes in multivariate data.
method A computationally feasible search method using random forests and class probability predictions.
result Consistently locates change points in simulations.
FJS method improves multinomial classification accuracy.
problem Improving multinomial classification accuracy under dataset shift.
method Derive FJS representation and propose alternative methods.
result Factorizable joint shift is not fully identifiable without additional assumptions.
In cases of uncertainty, a multi-class classifier preferably returns a set of candidate classes instead of predicting a single class label with little guarantee. More precisely, the classifier should strive for an optimal balance between the correctness (the true class is among the candidates) and the precision (the ca…
Flexible evidential deep learning improves uncertainty quantification in machine learning.
problem Overconfident predictions in machine learning models can lead to serious consequences.
method Proposes flexible evidential deep learning (F-EDL) to model uncertainty over class probabilities using a flexible Dirichlet distribution.
result Empirically demonstrates state-of-the-art uncertainty quantification performance across diverse scenarios.
New algorithms optimize metrics for binary classification with class imbalance.
problem Optimizing metrics like Fβ, AM, Jaccard for imbalanced classes.
method Reformulates metric optimization as cost-sensitive learning, using surrogate loss functions.
result METRO algorithms provide strong theoretical guarantees and outperform baselines.
A new metric for uncertainty quantification using class collisions.
problem Fine-grained uncertainty quantification in classification problems.
method Introducing the collision matrix and estimating it from one-hot labeled data.
result The collision matrix uniquely recovers the posterior class probability distribution.
Machine learning models are vulnerable to simple model stealing attacks if the adversary can obtain output labels for chosen inputs. To protect against these attacks, it has been proposed to limit the information provided to the adversary by omitting probability scores, significantly impacting the utility of the provid…
In binary classification framework, we are interested in making cost sensitive label predictions in the presence of uniform/symmetric label noise. We first observe that 0-1 Bayes classifiers are not (uniform) noise robust in cost sensitive setting. To circumvent this impossibility result, we present two schemes; un…
This paper develops convex surrogates for optimizing the multi-label F-measure.
problem Optimizing the F-measure for multi-label classification is computationally hard.
method Designing convex surrogate losses calibrated for the F-measure.
result The F-measure for multi-label problems has a rank of at most s2+1. A key prerequisite to optimal reasoning under uncertainty in intelligent systems is to start with good class probability estimates. This paper improves on the current best probability estimation trees (Bagged-PETs) and also presents a new ensemble-based algorithm (MOB-ESP). Comparisons are made using several benchmark …
This paper proposes an online knowledge distillation method that transfers feature map information in addition to class probabilities.
problem Previous online knowledge distillation methods only utilized class probabilities, missing feature map information.
method Adversarial training framework to transfer feature map information; multiple networks trained simultaneously with discriminators.
result Our method performs better than direct alignment methods and is more suitable for online distillation.
The paper addresses probability calibration for incomplete sequences.
problem Improving probability estimates from incomplete sequences.
method Adapting traditional calibration techniques to sequences of varying lengths.
result Proposed methods improve probability calibration for modern sequential models.
Deep learning models are vulnerable to external attacks. In this paper, we propose a Reinforcement Learning (RL) based approach to generate adversarial examples for the pre-trained (target) models. We assume a semi black-box setting where the only access an adversary has to the target model is the class probabilities o…
Assessing reliably the confidence of a deep neural network and predicting its failures is of primary importance for the practical deployment of these models. In this paper, we propose a new target criterion for model confidence, corresponding to the True Class Probability (TCP). We show how using the TCP is more suited…
We establish linear regret bounds for convex smooth losses using Fenchel-Young losses.
problem Establishing linear regret bounds for convex smooth losses.
method Constructing a convex smooth surrogate loss using Fenchel-Young losses generated by the convolutional negentropy.
result We derive a smooth loss with a linear surrogate regret bound.
Paper develops proper, lower-bounded losses for weakly supervised classification.
problem Weakly supervised classification with corrupted labels.
method Representation theorem for proper losses, derived condition for lower-boundedness, generalized logit squeezing.
result Proper and lower-bounded losses for weak-label learning.
Proposes novel wSVMs for sparse learning and accurate probability estimation.
problem Sparse features with redundant noise limit the performance of existing wSVMs.
method Develops ℓ1-norm and elastic net regularized wSVMs for automatic variable selection and probability estimation. result Elastic net regularized wSVMs achieve superior performance in variable selection and probability estimation.
New uniqueness concept for adversarial Bayes classifier.
problem Understanding adversarial Bayes classifiers in binary classification.
method Developed a new notion of uniqueness and analyzed it for a family of one-dimensional data distributions.
result Improved regularity of adversarial Bayes classifiers as perturbation radius increases.
Empirical Bayes rates via variational approximations and prior decomposition.
problem Nonparametric and high-dimensional inference convergence rates.
method Variational perspective and prior decomposition.
result Empirical Bayes posterior rates derived from variational Bayes.
The majority of traditional classification ru les minimizing the expected probability of error (0-1 loss) are inappropriate if the class probability distributions are ill-defined or impossible to estimate. We argue that in such cases class domains should be used instead of class distributions or densities to construct …
The study compares clustering risk in Hidden Markov and i.i.d. models, showing the Bayes classifier is nearly optimal.
problem Comparing clustering risk in Hidden Markov and i.i.d. models.
method Analysis of Bayes risk, theoretical bounds, and simulations.
result The Bayes classifier is nearly optimal for clustering in both Hidden Markov and i.i.d. models.
The article focuses on determining the predictive uncertainty of a model on the example of atrial fibrillation detection problem by a single-lead ECG signal. To this end, the model predicts parameters of the beta distribution over class probabilities instead of these probabilities themselves. It was shown that the desc…
Despite its simplicity, the naive Bayes classifier has surprised machine learning researchers by exhibiting good performance on a variety of learning problems. Encouraged by these results, researchers have looked to overcome naive Bayes primary weakness - attribute independence - and improve the performance of the algo…
A-GPS learns to generate Pareto sets efficiently with user preferences.
problem Online discrete multi-objective optimization with user preferences.
method Generative model with class probability estimator (CPE) for non-dominance and preference alignment.
result Amortized generative model for efficient Pareto set approximation.
KCal calibrates deep networks by embedding logits in a metric space.
problem Overconfident predictions from DNNs, especially in high-risk applications.
method KCal learns a metric space on the penultimate-layer latent embedding and generates predictions using kernel density estimates.
result KCal provides a provable full calibration guarantee and consistently outperforms baselines.
Researchers prove NP-hardness of learning parameter-bounded Bayes nets.
problem Learning parameter-bounded Bayes nets is computationally hard.
method Proved NP-hardness of learning parameter-bounded Bayes nets and a promise search variant.
result Proved NP-hardness of a promise search variant of LEARN.
Deep learning improves Bayes factor computation for likelihood-free models.
problem Computing Bayes factors for likelihood-free models is challenging.
method Proposes a deep learning estimator of Bayes factors using simulated data.
result Establishes consistency of the Deep Bayes Factor estimator.
Meta-learning bounds derived using PAC-Bayes theory for improved generalization.
problem Uncertainty in generalization performance for meta-learning with new tasks.
method PAC-Bayes relative entropy bounds and empirical risk minimization (ERM) method.
result Competitive generalization performance and rapid convergence with data-dependent prior.
We derive the fast convergence rates of a deep neural network (DNN) classifier with the rectified linear unit (ReLU) activation function learned using the hinge loss. We consider three cases for a true model: (1) a smooth decision boundary, (2) smooth conditional class probability, and (3) the margin condition (i.e., t…
PAC-Bayes framework fails on simple 1D linear classification task.
problem Proving the learnability of simple 1D linear classification tasks using PAC-Bayes bounds.
method Demonstrated a specific 1D linear classification task that PAC-Bayes cannot analyze.
result PAC-Bayes framework cannot prove learnability of simple 1D linear classification tasks.
Improves GAN-based semi-supervised learning with consistency regularization.
problem Lack of consistency in class probability predictions under local perturbations.
method Introduces consistency regularization to GANs, leveraging both local and interpolation consistency.
result Significantly improves performance and achieves new state-of-the-art results.
PAC-Bayes bound requires prior to place mass on high-performing predictors.
problem Explaining generalization in machine learning.
method Analyzing necessary conditions for PAC-Bayes bounds to provide meaningful generalization guarantees.
result Achieving a target generalisation level requires the prior to place sufficient mass on high-performing predictors.
New estimators outperform maximum likelihood without hyper-parameter estimation.
problem Improving system identification performance without hyper-parameter estimation.
method Developed generalized Bayes and closed-form biased estimators using excess MSE.
result New estimators have comparable performance to empirical-Bayes-based regularized estimator.
Paper introduces Generalized Naive Bayes for better data fitting.
problem Improving Naive Bayes for better data fitting.
method Developed new greedy and optimal algorithms for GNB.
result Proves GNB fits data at least as well as classical NB.
Smart Bayes integrates generative and discriminative features for improved classification.
problem Improving classification performance by combining generative and discriminative modeling.
method Integrates generative likelihood-ratio features into a logistic-regression-style classifier.
result Often outperforms logistic regression and Naive Bayes in simulations and real data.
PFNs pre-train models on simulated data to predict class probabilities.
problem Training machine learning models on large datasets.
method Pre-train a fixed model on small simulated datasets and use it to infer class probabilities in-context.
result PFNs achieve state-of-the-art performance and improve with larger inference data.
As part of autonomous car driving systems, semantic segmentation is an essential component to obtain a full understanding of the car's environment. One difficulty, that occurs while training neural networks for this purpose, is class imbalance of training data. Consequently, a neural network trained on unbalanced data …
New method reduces computational cost for estimating PAC-Bayes bounds.
problem High computational cost in estimating PAC-Bayes bounds.
method General alternative method that makes computational savings.
result Reduces computational cost on the order of the dataset size.