Feature-based time series classification (TSC) methods have traditionally relied on time-domain features, which can limit their effectiveness in capturing the full spectrum of temporal dynamics. This study introduces Crossfire (Cross-representation Feature Integration for Robust Extraction, CFIRE), a feature-based framework that systematically extracts and integrates features from a diverse set of time series representations—including derivative, autocorrelation, spectral (Fourier), harmonic (cosine), time-frequency (wavelet), and analytic (Hilbert) domains. The pipeline supports parallelized feature extraction and incorporates selection mechanisms to manage redundancy while enhancing classification performance. Empirical evaluation across all 142 datasets in the UCR archive demonstrates that CFIRE outperforms leading feature-based baselines such as FreshPRINCE. Although it does not always surpass state-of-the-art (SOTA) accuracy in other TSC paradigms, CFIRE provides distinctive advantages in computational efficiency, interpretability, and scalability. Notably, its computational efficiency is comparable to that of Quant, one of the fastest existing TSC methods, while maintaining robust accuracy across longer sequences and larger class counts. The results indicate that the combination of features across multiple representations—when systematically integrated and optimized—can provide a practical, scalable, and interpretable alternative to more complex TSC approaches. The full implementation is publicly available at: https://doi.org/10.5281/zenodo.15695653 .

Human Activity Recognition (HAR) has been extensively studied, with recent emphasis on the implementation of advanced Machine Learning (ML) and Deep Learning (DL) algorithms for accurate classification. This study investigates the efficacy of two ML algorithms, eXtreme Gradient Boosting (XGBoost) and MiniRocket, in the realm of HAR using data collected from smartphone sensors. The experiments are conducted on a dataset obtained from the UCI repository, comprising accelerometer and gyroscope signals captured from 30 volunteers performing various activities while wearing a smartphone. The dataset undergoes preprocessing, including noise filtering and feature extraction, before being utilized for training and testing the classifiers. Monte Carlo cross-validation is employed to evaluate the models' robustness. The findings reveal that both XGBoost and MiniRocket attain accuracy, F1 score, and AUC values as high as 0.99 in activity classification. XGBoost exhibits a slightly superior performance compared to MiniRocket. Notably, both algorithms surpass the performance of other ML and DL algorithms reported in the literature for HAR tasks. Additionally, the study compares the computational efficiency of the two algorithms, revealing XGBoost's advantage in terms of training time. Furthermore, the performance of MiniRocket, which achieves accuracy and F1 values of 0.94, and an AUC value of 0.96 using raw data and utilizing only one channel from the sensors, highlights the potential of directly leveraging unprocessed signals. It also suggests potential advantages that could be gained by utilizing sensor fusion or channel fusion techniques. Overall, this research sheds light on the effectiveness and computational characteristics of XGBoost and MiniRocket in HAR tasks, providing insights for future studies in activity recognition using smartphone sensor data.

Atrial Fibrillation (AF) detection from intracardiac Electrogram (EGM) signals is a critical aspect of cardiovascular health monitoring. This study explores the application of Minirocket, a time series classification (TSC) algorithm, for robust and efficient AF detection. A comparative analysis is conducted against a deep learning approach using a subset of the dataset from Rodrigo et al. (2022). The study investigates the robustness of Minirocket in the face of shorter EGM sequences and varying training sizes, essential for real-world applications such as wearable and implanted devices. Empirical runtime analysis further assesses the efficiency of Minirocket in comparison to conventional machine learning (ML) algorithms. The results showcase Minirocket's notable performance, especially in scenarios with shorter signals and varying training sizes, making it a promising candidate for streamlined AF detection in emerging cardiovascular monitoring technologies. This research contributes to the optimization of AF detection algorithms for increased efficiency and adaptability to dynamic clinical scenarios.

Hierarchical classification (HC) plays a pivotal role in multi-class classification tasks, where objects are organized into a hierarchical structure. This study explores the performance of HC through a comprehensive analysis that encompasses both hierarchy generation and hierarchy exploitation. This analysis is particularly relevant in scenarios where a predefined hierarchy structure is not readily accessible. Notably, two novel hierarchy exploitation schemes, LCPN+ and LCPN+F, which extend the capabilities of LCPN and combine the strengths of global and local classification, have been introduced and evaluated alongside existing methods. The findings reveal the consistent superiority of LCPN+F, which outperforms other schemes across various datasets and scenarios. Moreover, this research emphasizes not only effectiveness but also efficiency, as LCPN+ and LCPN+F maintain runtime performance comparable to Flat Classification (FC). Additionally, this study underscores the importance of selecting the right hierarchy exploitation scheme to maximize classification performance. This work extends our understanding of HC and establishes a benchmark for future research, fostering advancements in multi-class classification methodologies.