Abstract

Abstract

Abstract
Effective selection of forecasting algorithms for time series data is a challenge in machine learning, impacting both predictive accuracy and efficiency. Metalearning, using features extracted from time series, offers a strategic approach to optimize algorithm selection. The utility of this approach depends on the amount of information the features contain about the behavior of the algorithms. Although there are several methods for systematic time series feature extraction, they have never been compared. This paper empirically analyzes the performance of each feature extraction method for algorithm selection and its impact on forecasting accuracy. Our study reveals that TSFRESH, TSFEATURES, and TSFEL exhibit comparable performance at algorithm selection accuracy, adeptly capturing time series characteristics essential for accurate algorithm selection. In contrast, Catch22 is found to be less effective for this purpose. In particular, TSFEL is identified as the most efficient method, balancing dimensionality and predictive performance. These findings provide insights for enhancing forecasting accuracy and efficiency through judicious selection of meta-feature extractors.

Abstract
Predictive Maintenance (PdM) aims to prevent failures through early detection, yet lacks explainability to support decision-making. Current PdM models often identify failures, but fail to explain their root causes, especially in real-world scenarios, with complex and limited labeled data. This study proposes an interpretable framework that combines LSTM-based Anomaly Detection with a dual-layered Root Cause Analysis (RCA) based on SHAP attributions. Applied to a real-world dataset, the method detects degradation transitions, tracks failure patterns over time, and provides interpretable information without explicit root cause labels.

Abstract

Abstract
The goal of providing legal identity to all individuals remains a global challenge. Current digital identity models depend on the existence of authoritative issuers, which are not always available, particularly in Low-Income Countries (LICs) and Lower-Middle-Income Countries (LMICs). This limitation severely restricts the coverage of identity systems. In response, new approaches such as the non-authoritative identity model have emerged, enabling entities with local reputation to serve as primary sources of information about individuals. However, existing literature does not yet offer a comprehensive solution for implementing such systems. In this paper, we present Nexus, a data fusion module that enables reputable entities to serve as primary providers of identity information, facilitating the design and implementation of digital identity systems that complement the State or other centralized authorities. A central element of the proposed data fusion component is a novel scheme combining truth discovery algorithms and reputation tracking. The feasibility, behavior, and performance of the proposed approach are demonstrated through experiments conducted on multiple datasets. In addition, Nexus was evaluated using a dataset from a health line service in a LIC, thereby demonstrating its applicability in real-world settings. This work represents an essential step toward eliminating barriers to pursuing a comprehensive legal identity.