Time series anomaly detection has received substantial attention over the past two decades, leading to the development of hundreds of algorithms. However, comprehensively understanding this vast landscape remains challenging, particularly for non-experts and novices. In this demonstration paper, we present InTimeAD, an interactive web application that provides access to more than 30 state-of-the-art time series anomaly detection algorithms. InTimeAD is intended to explore the performance of existing as well as custom anomaly detection models in an interactive, hands-on manner. By lowering the entry bar, we support practitioners overwhelmed by the large number of existing techniques, while providing a platform for researchers to rapidly analyze their novel anomaly detection algorithms.

Joint-Embedding Predictive Architectures (JEPAs), a powerful class of self-supervised models, exhibit an unexplained ability to cluster time-series data by their underlying dynamical regimes. We propose a novel theoretical explanation for this phenomenon, hypothesizing that JEPA's predictive objective implicitly drives it to learn the invariant subspace of the system's Koopman operator. We prove that an idealized JEPA loss is minimized when the encoder represents the system's regime indicator functions, which are Koopman eigenfunctions. This theory was validated on synthetic data with known dynamics, demonstrating that constraining the JEPA's linear predictor to be a near-identity operator is the key inductive bias that forces the encoder to learn these invariants. We further discuss that this constraint is critical for selecting this interpretable solution from a class of mathematically equivalent but entangled optima, revealing the predictor's role in representation disentanglement. This work demystifies a key behavior of JEPAs, provides a principled connection between modern self-supervised learning and dynamical systems theory, and informs the design of more robust and interpretable time-series models.

Reliable anomaly detection under data contamination remains a major challenge in Prognostics and Health Management, especially when degradation processes are gradual and clean training data are unavailable. This paper introduces a weighted iterative refinement framework with autoencoders for contaminated anomaly detection (WIRACAD) to address this problem. The method, which is based on reconstruction residuals, re-weights training samples across iterations in order to progressively reduce the influence of suspected anomalies. This continuous refinement improves the robustness of health indicator learning from contaminated time series. The proposed approach is evaluated on two public benchmarks: the NASA C-MAPSS dataset and the IMS Bearing dataset. Results show consistent improvements in key metrics related to degradation monitoring. In particular, the overal fit score and the monotonicty are improved when compared to baseline autoencoder training and recent refinement-based methods. These findings suggest that iterative sample weighting can enhance unsupervised anomaly detection with autoencoders in settings where data contamination is assumed.

Time series semantic segmentation is the task of extracting time intervals from the time series data that share a similar meaning within the application domain in an unsupervised manner. State-of-the-art algorithms typically treat this problem as change point detection, resulting in discrete state transitions. However, in real-world applications, states often transition gradually. This leads to a novel, more challenging variation of the traditional time series segmentation task, for which we present PaTSS, a novel, domain-agnostic algorithm to uncover those gradual state transitions. PaTSS learns a distribution over the semantic segments based on an embedding space derived from mined sequential patterns. An extensive experimental evaluation on 107 benchmark time series shows that PaTSS is capable of detecting gradual state transitions, a task current methods are unable to perform.

In the food industry, an accurate determination of the quantity of ingredients is of key importance. On the one hand, because this is prescribed by European regulations, which require a precise quantity indication on the food labeling. On the other hand, because small variations in the fat quantity and type produce a different taste experience for the consumer. Since current automated production lines for bakery products often cause fluctuations in the fat concentration from the target value, accurate and timely ingredient concentration measurements are required for quality control. However, conventional wet chemistry-based methods are a manual, destructive form of testing that require a considerable amount of time, typically several hours to days, to generate an outcome. To this end, this paper investigates the potential of machine learning-based regression using features extracted from hyperspectral imaging for the non-invasive and inline prediction of fat and moisture concentration in bakery products. While the conventional testing method is still advantageous in terms of accuracy, the results show that the proposed hyperspectral imaging approach is a promising alternative due to its real-time and non-destructive nature, offering the possibility to inspect larger quantities of products.

Regularization is one of the key concepts in machine learning, but so far it has received only little attention in the logical and relational learning setting. Here we propose a regularization and feature selection technique for such setting, in which one commonly represents the structure of the domain using an entity-relationship model. To this end, we introduce a notion of locality that ties together features according to their proximity in a transformed representation of the relational learning problem obtained via a procedure that we call “graphicalization”. We present two techniques, a wrapper and an efficient embedded approach, to identify the most relevant sets of predicates which yields more readily interpretable results than selecting low-level propositionalized features. The proposed techniques are implemented in the kernel-based relational learner kLog, although the ideas presented here can also be adapted to other relational learning frameworks. We evaluate our approach on classification tasks in the natural language processing and bioinformatics domain.