Found 2674 publications. Showing page 9 of 268:
Stochastic and deterministic processes in Asymmetric Tsetlin Machine
This paper introduces a new approach to enhance the decision-making capabilities of the Tsetlin Machine (TM) through the Stochastic Point Location (SPL) algorithm and the Asymmetric Steps technique. We incorporate stochasticity and asymmetry into the TM's process, along with a decaying normal distribution function that improves adaptability as it converges toward zero over time. We present two methods: the Asymmetric Probabilistic Tsetlin (APT) Machine, influenced by random events, and the Asymmetric Tsetlin (AT) Machine, which transitions from probabilistic to deterministic states. We evaluate these methods against traditional machine learning algorithms and classical Tsetlin (CT) machines across various benchmark datasets. Both AT and APT demonstrate competitive performance, with the AT model notably excelling, especially in complex datasets.
2025
Stress is a common human reaction to demanding circumstances, and prolonged and excessive stress can have detrimental effects on both mental and physical health. Heart rate variability (HRV) is widely used as a measure of stress due to its ability to capture variations in the time intervals between heartbeats. However, achieving high accuracy in stress detection through machine learning (ML), using a reduced set of statistical features extracted from HRV, remains a significant challenge. In this study, we aim to address these challenges by proposing lightweight ML models that can effectively detect stress using minimal HRV features and are computationally efficient enough for IoT deployment. We have developed ML models incorporating efficient feature selection techniques and hyper-parameter tuning. The publicly available SWELL-KW dataset has been utilized for evaluating the performance of our models. Our results demonstrate that lightweight models such as k-NN and Decision Tree can achieve competitive accuracy while ensuring lower computational demands, making them ideal for real-time applications. Promisingly, among the developed models, the k-nearest neighbors (k-NN) algorithm has emerged as the best-performing model, achieving an accuracy score of 99.3% using only three selected features. To confirm real-world deployability, we benchmarked the best model on an 8 GB NVIDIA Jetson Orin Nano edge device, where it retained 99.26% accuracy and completed training in 31 s. Furthermore, our study has incorporated local interpretable model-agnostic explanations to provide comprehensive insights into the predictions made by the k-NN-based architecture.
2025
2025
2025
2025
Supervised Anomaly Detection in Univariate Time-Series Using 1D Convolutional Siamese Networks
In time-series data analysis, identifying anomalies is crucial for maintaining data integrity and ensuring accurate analyses and decision-making. Anomalies can compromise data quality and operational efficiency. The complexity of time-series data, with its temporal dependencies and potential non-stationarity, makes anomaly detection challenging but essential. Our research introduces ADSiamNet, a 1D Convolutional Neural Network-based Siamese network model for anomaly detection and rectification. ADSiamNet effectively identifies localized patterns in time-series data and smooths detected anomalies using a quantile-based technique. In tests with physical activity data from Actigraph watches and MOX2-5 sensors, ADSiamNet achieved accuracies of 98.65% and 85.0%, respectively, outperforming other supervised anomaly detection methods. The model uses a contrastive loss function to compare input sequences and adjusts network weights iteratively during training to recognize intricate patterns. Additionally, we evaluated various univariate time-series forecasting algorithms on datasets with and without anomalies. Results show that anomaly-smoothed data reduces forecasting errors, highlighting our approach’s effectiveness in enhancing time-series data analysis’s integrity and reliability. Future research will focus on multivariate time-series datasets.
2025
Are ingredients of personal care products likely to undergo long-range transport to remote regions?
Personal care products (PCPs) contain contaminants of emerging concern. Despite increasing reports of their presence in polar regions, the behavior of PCP ingredients under cold environmental conditions remains poorly understood. Snow collected around Villum Research Station at Station Nord, Greenland, between December 2018 and June 2019 was extracted in a stainless steel clean-room and analyzed for seven fragrance materials, four organic UV-filters and an antioxidant using gas chromatography-tandem mass spectrometry. All twelve target PCPs were detected, with elevated concentrations during two sampling events potentially tied to air mass transport from northern Europe and the northern coasts of Russia. To contextualize the presence of these PCP chemicals in high Arctic snow, we estimated their (i) partitioning properties as a function of temperature, (ii) equilibrium phase distribution and dominant deposition processes in the atmosphere at temperatures above and below freezing, and (iii) potential for long-range environmental transport (LRET). Even though most PCPs are deemed to be gas phase chemicals predominantly deposited as vapors, rapid atmospheric degradation is expected to limit their LRET. On the other hand, the less volatile octocrylene is expected to be sorbed to atmospheric particles, removed via wet and dry particle deposition, and possibly exhibit a higher potential for LRET by being protected from attack by photooxidants. The contrast between consistent detection of PCP chemicals in high Arctic snow and relatively low estimated LRET potential emphasizes the need for further research on their real-world atmospheric behavior under cold conditions.
2025
The apportionment of equivalent black carbon (eBC) to combustion sources from liquid fuels (mainly fossil; eBCLF) and solid fuels (mainly non-fossil; eBCSF) is commonly performed using data from Aethalometer instruments (AE approach). This study evaluates the feasibility of using AE data to determine the absorption Ångström exponents (AAEs) for liquid fuels (AAELF) and solid fuels (AAESF), which are fundamental parameters in the AE approach. AAEs were derived from Aethalometer data as the fit in a logarithmic space of the six absorption coefficients (470–950 nm) versus the corresponding wavelengths. The findings indicate that AAELF can be robustly determined as the 1st percentile (PC1) of AAE values from fits with R2 > 0.99. This R2-filtering was necessary to remove extremely low and noisy-driven AAE values commonly observed under clean atmospheric conditions (i.e., low absorption coefficients). Conversely, AAESF can be obtained from the 99th percentile (PC99) of unfiltered AAE values. To optimize the signal from solid fuel sources, winter data should be used to calculate PC99, whereas summer data should be employed for calculating PC1 to maximize the signal from liquid fuel sources. The derived PC1 (AAELF) and PC99 (AAESF) values ranged from 0.79 to 1.08, and 1.45 to 1.84, respectively. The AAESF values were further compared with those constrained using the signal at mass-to-charge 60 (m/z 60), a tracer for fresh biomass combustion, measured using aerosol chemical speciation monitor (ACSM) and aerosol mass spectrometry (AMS) instruments deployed at 16 sites. Overall, the AAESF values obtained from the two methods showed strong agreement, with a coefficient of determination (R2) of 0.78. However, uncertainties in both approaches may vary due to site-specific sources, and in certain environments, such as traffic-dominated sites, neither approach may be fully applicable.
2025
Recommendation technologies are widespread in streaming services, e-commerce, social media, news, and content management. Besides recommendation generation, its presentation is also important. Most research and development focus on the technical aspects of recommendation generation; therefore, a gap exists between recommendation generation and its effective presentation and user interaction. This study focuses on how personalized recommendations can be presented and interacted with in a music recommendation system using interactive visual interfaces. Interactive interface modeling with User-Centered Design (UCD) in a recommendation system is essential for creating a user-friendly, engaging, and personalized experience. By involving users in the recommendation process and considering their feedback, the system can deliver more relevant content, foster user trust, and improve overall user satisfaction and engagement. In this study, the visual interface design and development of a personalized music recommendation prototype (MusicReco) are presented using an iterative UCD approach, involving twenty end-users, one researcher, three academic professionals, and four experts. As the study is more inclined toward the recommendation presentation and visual modeling, we used a standard content-based filtering algorithm on the publicly available Spotify dataset for music recommendation generation. End-users helped to mature the MusicReco prototype to a basic working version through continuous feedback and design inputs on their needs, context, preferences, personalization, and effective visualization. Moreover, MusicReco captures the idea of mood-based tailored recommendations to encourage end-users. Overall, this study demonstrates how UCD can enhance the presentation and interaction of mood-based music recommendations, effectively engaging users with advancements in recommendation algorithms as a future focus.
2025
Stress management with HRV following AI, semantic ontology, genetic algorithm and tree explainer
Heart Rate Variability (HRV) serves as a vital marker of stress levels, with lower HRV indicating higher stress. It measures the variation in the time between heartbeats and offers insights into health. Artificial intelligence (AI) research aims to use HRV data for accurate stress level classification, aiding early detection and well-being approaches. This study’s objective is to create a semantic model of HRV features in a knowledge graph and develop an accurate, reliable, explainable, and ethical AI model for predictive HRV analysis. The SWELL-KW dataset, containing labeled HRV data for stress conditions, is examined. Various techniques like feature selection and dimensionality reduction are explored to improve classification accuracy while minimizing bias. Different machine learning (ML) algorithms, including traditional and ensemble methods, are employed for analyzing both imbalanced and balanced HRV datasets. To address imbalances, various data formats and oversampling techniques such as SMOTE and ADASYN are experimented with. Additionally, a Tree-Explainer, specifically SHAP, is used to interpret and explain the models’ classifications. The combination of genetic algorithm-based feature selection and classification using a Random Forest Classifier yields effective results for both imbalanced and balanced datasets, especially in analyzing non-linear HRV features. These optimized features play a crucial role in developing a stress management system within a Semantic framework. Introducing domain ontology enhances data representation and knowledge acquisition. The consistency and reliability of the Ontology model are assessed using Hermit reasoners, with reasoning time as a performance measure. HRV serves as a significant indicator of stress, offering insights into its correlation with mental well-being. While HRV is non-invasive, its interpretation must integrate other stress assessments for a holistic understanding of an individual’s stress response. Monitoring HRV can help evaluate stress management strategies and interventions, aiding individuals in maintaining well-being.
2025