An end-to-end distribution- and task-aware parallel imputation method for improving target task performance

Adaptive Bayesian Early-Exit Networks for Efficient Non-Transferable Learning

Labeled-Data-Free Meta-Learning: Efficient Task Generation Using Pre-trained Models and Unlabeled Data

Meta-learning without labeled data is crucial for real-world applications, where obtaining labeled datasets can be expensive or restricted due to privacy concerns. Data-Free Meta-Learning (DFML) addresses this challenge by leveraging pre-trained models without access to training data. However, existing DFML methods rely on model inversion to generate training data, a process that is generally difficult and computationally expensive due to the need to generate high-dimensional data matching the original distribution. To address this limitation, we propose a novel meta-learning setting that avoids model inversion by jointly leveraging pre-trained models and unlabeled data. Our method generates meta-training tasks by assigning soft labels from pre-trained models to unlabeled data. Since the quality of these tasks can vary, we introduce a task-weighting mechanism based on task confidence and class distribution balance to ensure effective meta-learning. Extensive experiments demonstrate that our approach substantially reduces computational cost and improves generalization, achieving up to 104-fold speedup and 8.4 percent to 36.4 percent improvements in few-shot classification accuracy compared to state-of-the-art DFML methods.

Pitfalls and Remedies for Multi-Task Bayesian Optimization

Bayesian optimization routinely warm-starts a target experiment with data from related source tasks, and the multi-task Gaussian process is the textbook surrogate for the job. We revisit this default in a controlled setting and find that it misestimates the cross-task correlation even in the simplest non-trivial case, affinely related source and target tasks, where a working transfer learning method should obviously succeed. We trace the failure to two independent structural mechanisms. Per-task standardization, the textbook fix for the affine slice ambiguity, propagates a finite-sample alignment error into the recovered correlation. The marginal likelihood itself identifies the correlation only at a per-sample rate that a Gaussian process at non-overlapping designs further dilutes. We propose three conservative remedies that follow from the analysis: promoting per-task means and scales to model parameters, restricting the task covariance to non-negative correlations, and co-locating part of the source and target designs. Across synthetic multi-task problems and surrogate-based hyperparameter tuning transfer, these remedies recover the target-only baseline on the simple instances, while the broader failure persists on harder instances and across most rank-based and latent-context variants.

Learning the Pareto Frontier of Predictive Models under Distribution Shift

Modern machine learning pipelines increasingly rely on reusing pretrained and foundation models across downstream tasks. These pretrained models can differ not only in performance but also in how they can be used: some only provide black-box predictions, while others may permit white-box access to internal representations that can be probed or fine-tuned. When deployed to the target domain in the presence of distribution shift, no single strategy, including zero-shot application, fine-tuning, or directly training a target-specific model, is uniformly the best. In this work, we propose Frontier Learning, a framework that treats a library of candidate models spanning different training histories and access regimes as complementary sources of information rather than mutually exclusive alternatives. Frontier Learning constructs a unified target-domain feature by concatenating internal representations from white-box candidates as well as prediction outputs from black-box candidates, then fits a lightweight, regularized supervised learner on this concatenated representation using labeled target data. Because the resulting hypothesis class contains predictors obtained by zero-shot reuse, fine-tuning, and direct training as special cases, empirical risk minimization over the frontier learner is guaranteed to be no worse, on the training sample, than any individual baseline. We evaluate the framework in simulations spanning varying degrees of source-target compatibility and in two real-world distribution-shift settings: visual domain adaptation on DomainNet/VisDA and clinical mortality prediction across intensive care unit domains using MIMIC-IV-Notes. Across all settings, Frontier Learning matches or outperforms the strongest individual reuse strategy, with the largest gains arising precisely when no single baseline is reliable across the range of shift considered.

Emputation: Identification-Guided Neural Imputation Framework

We propose Emputation, a deep generative framework for learning imputation models. Emputation targets the extrapolation distribution of missing variables given observed variables, and training is guided by specific missingness assumptions that guarantee identification of the target distribution. The training objective, called the emputation risk, is an energy-score-based risk in which the identification assumption determines how observed entries are masked and which observations contribute to training. The resulting framework enables direct conditional sampling for multiple imputation. We show that the population minimizer of the emputation risk recovers the target extrapolation distribution under a broad class of identification assumptions, including several missing-not-at-random assumptions. Simulations show strong performance under both pointwise and distributional evaluation metrics, and an application to an Alzheimer's disease dataset demonstrates its practical value.

Not All Missing Data are Equal: Choosing the Right Imputation Method for Binary Datasets

Missing binary predictors are common in reliability, quality control, and industrial decision systems, yet imputation methods are often chosen by convenience rather than evidence. We conduct a Monte Carlo study comparing mode substitution, sequential hot‐deck, missForest, MICE, and KNN with three neighbourhood sizes under MCAR, MAR, and MNAR missingness, across missingness rates from 5% to 50% and two predictor‐dependence structures. Performance is evaluated on three targets: exact recovery of missing binary cells, recovery of logistic‐regression coefficients, and downstream classification using logistic regression, naive Bayes, support vector machines, and random forests. The results reveal a clear trade‐off. KNN is strongest for exact cell recovery under MCAR and MAR, whereas missForest performs best under MNAR. MICE is the most reliable choice for downstream predictive performance across learners and missingness mechanisms. By contrast, mode imputation and sequential hot‐deck achieve the best coefficient recovery. The main implication is operational: in binary‐data environments, imputation should be chosen to match the analytical objective–reconstruction, inference, or prediction–because no single method dominates all targets simultaneously.