Conformal prediction constructs a set of labels instead of a single point prediction, while providing a probabilistic coverage guarantee. Beyond the coverage guarantee, adaptiveness to example difficulty is an important property. It means that the method should produce larger prediction sets for more difficult examples, and smaller ones for easier examples. Existing evaluation methods for adaptiveness typically analyze coverage rate violation or average set size across bins of examples grouped by difficulty. However, these approaches often suffer from imbalanced binning, which can lead to inaccurate estimates of coverage or set size. To address this issue, we propose a binning method that leverages input transformations to sort examples by difficulty, followed by uniform-mass binning. Building on this binning, we introduce two metrics to better evaluate adaptiveness. These metrics provide more reliable estimates of coverage rate violation and average set size due to balanced binning, leading to more accurate adaptivity assessment. Through experiments, we demonstrate that our proposed metric correlates more strongly with the desired adaptiveness property compared to existing ones. Furthermore, motivated by our findings, we propose a new adaptive prediction set algorithm that groups examples by estimated difficulty and applies group-conditional conformal prediction. This allows us to determine appropriate thresholds for each group. Experimental results on both (a) an Image Classification (ImageNet) (b) a medical task (visual acuity prediction) show that our method outperforms existing approaches according to the new metrics.

Recent research on time series foundation models has primarily focused on forecasting, leaving it unclear how generalizable their learned representations are. In this study, we examine whether frozen pre-trained forecasting models can provide effective representations for classification. To this end, we compare different representation extraction strategies and introduce two model-agnostic embedding augmentations. Our experiments show that the best forecasting models achieve classification accuracy that matches or even surpasses that of state-of-the-art models pre-trained specifically for classification. Moreover, we observe a positive correlation between forecasting and classification performance. These findings challenge the assumption that task-specific pre-training is necessary, and suggest that learning to forecast may provide a powerful route toward constructing general-purpose time series foundation models.

Deep learning dominates image classification tasks, yet understanding how models arrive at predictions remains a challenge. Much research focuses on local explanations of individual predictions, such as saliency maps, which visualise the influence of specific pixels on a model's prediction. However, reviewing many of these explanations to identify recurring patterns is infeasible, while global methods often oversimplify and miss important local behaviours. To address this, we propose Segment Attribution Tables (SA Ts), a method for summarising local saliency explanations into (semi-)global insights. SA Ts take image segments (such as "eyes" in Chihuahuas) and leverage saliency maps to quantify their influence. These segments highlight concepts the model relies on across instances and reveal spurious correlations, such as reliance on backgrounds or watermarks, even when out-of-distribution test performance sees little change. SA Ts can explain any classifier for which a form of saliency map can be produced, using segmentation maps that provide named segments. SA Ts bridge the gap between oversimplified global summaries and overly detailed local explanations, offering a practical tool for analysing and debugging image classifiers.

The popularity of deep learning methods in the time series domain boosts interest in interpretability studies, including counterfactual (CF) methods. CF methods identify minimal changes in instances to alter the model predictions. Despite extensive research, no existing work benchmarks CF methods in the time series domain. Additionally, the results reported in the literature are inconclusive due to the limited number of datasets and inadequate metrics. In this work, we redesign quantitative metrics to accurately capture desirable characteristics in CFs. We specifically redesign the metrics for sparsity and plausibility and introduce a new metric for consistency. Combined with validity, generation time, and proximity, we form a comprehensive metric set. We systematically benchmark 6 different CF methods on 20 univariate datasets and 10 multivariate datasets with 3 different classifiers. Results indicate that the performance of CF methods varies across metrics and among different models. Finally, we provide case studies and a guideline for practical usage.

Dynamic Time Warping (DTW) is a popular time series distance measure that aligns the points in two series with one another. These alignments support warping of the time dimension to allow for processes that unfold at differing rates. The distance is the minimum sum of costs of the resulting alignments over any allowable warping of the time dimension. The cost of an alignment of two points is a function of the difference in the values of those points. The original cost function was the absolute value of this difference. Other cost functions have been proposed. A popular alternative is the square of the difference. However, to our knowledge, this is the first investigation of both the relative impacts of using different cost functions and the potential to tune cost functions to different tasks. We do so in this paper by using a tunable cost function {\lambda}{\gamma} with parameter {\gamma}. We show that higher values of {\gamma} place greater weight on larger pairwise differences, while lower values place greater weight on smaller pairwise differences. We demonstrate that training {\gamma} significantly improves the accuracy of both the DTW nearest neighbor and Proximity Forest classifiers.

Many deep models have been recently proposed for anomaly detection. This paper presents comparison of selected generative deep models and classical anomaly detection methods on an extensive number of non--image benchmark datasets. We provide statistical comparison of the selected models, in many configurations, architectures and hyperparamaters. We arrive to conclusion that performance of the generative models is determined by the process of selection of their hyperparameters. Specifically, performance of the deep generative models deteriorates with decreasing amount of anomalous samples used in hyperparameter selection. In practical scenarios of anomaly detection, none of the deep generative models systematically outperforms the kNN.

There are now a broad range of time series classification (TSC) algorithms designed to exploit different representations of the data. These have been evaluated on a range of problems hosted at the UCR-UEA TSC Archive (www.timeseriesclassification.com), and there have been extensive comparative studies. However, our understanding of why one algorithm outperforms another is still anecdotal at best. This series of experiments is meant to help provide insights into what sort of discriminatory features in the data lead one set of algorithms that exploit a particular representation to be better than other algorithms. We categorise five different feature spaces exploited by TSC algorithms then design data simulators to generate randomised data from each representation. We describe what results we expected from each class of algorithm and data representation, then observe whether these prior beliefs are supported by the experimental evidence. We provide an open source implementation of all the simulators to allow for the controlled testing of hypotheses relating to classifier performance on different data representations. We identify many surprising results that confounded our expectations, and use these results to highlight how an over simplified view of classifier structure can often lead to erroneous prior beliefs. We believe ensembling can often overcome prior bias, and our results support the belief by showing that the ensemble approach adopted by the Hierarchical Collective of Transform based Ensembles (HIVE-COTE) is significantly better than the alternatives when the data representation is unknown, and is significantly better than, or not significantly significantly better than, or not significantly worse than, the best other approach on three out of five of the individual simulators.