Carlos Manuel Soares

Abstract
When the available data for a target domain is limited, transfer learning (TL) methods leverage related data-rich source domains to train and evaluate models, before deploying them on the target domain. However, most TL methods assume fixed levels of labeled and unlabeled target data, which contrasts with real-world scenarios where both data and labels arrive progressively over time. As a result, evaluations based on these static assumptions may not reflect how methods perform in practice. To support a more realistic assessment of TL methods in dynamic settings, we propose an evaluation framework that (1) simulates varying data availability over time, (2) creates multiple domains via resampling of a given dataset and (3) introduces inter-domain variability through controlled transformations, e.g., including time-dependent covariate and concept shifts. These capabilities enable the systematic simulation of a large number of variants of the experiments, providing deeper insights into how algorithms may behave when deployed. We demonstrate the usefulness of the proposed framework by performing a case study on a proprietary real-world suite of card payment datasets. To support reproducibility, we also apply the framework on the publicly available Bank Account Fraud (BAF) dataset. By providing a methodology for evaluating TL methods over time and in different data availability conditions, our framework supports a better understanding of model behavior in real-world environments, which enables more informed decisions when deploying models in new domains.

Abstract

Abstract

Abstract

Abstract
Time series forecasting is pivotal across industries, as it fosters data-driven decision-making, increasing the chances of successful outcomes. Yet, certain instances that feature adverse characteristics, may lead models to manifest stress through decreases in performance (e.g., large errors). Hence, the ability to preemptively identify such cases, while establishing their root causes, would be advantageous to elevate the understanding of forecasting processes, informing users about the trustworthiness of predictions. Hence, we propose MASTFM, a method based on meta-learning that leverages statistical characteristics of input time series, and estimations of forecasting performance from model outputs, to build a metamodel that learns conditions for stress. Given that such occurrences are naturally rare, data augmentation is employed to ensure balance during training. Moreover, SHapley Additive exPlanations (SHAP) are used to explain how features impact forecasting behaviour.