Effect of Dataset Size and Train/Test Split Ratios in QSAR/QSPR Multiclass Classification

Molecules, Journal Year: 2021, Volume and Issue: 26(4), P. 1111 - 1111

Published: Feb. 19, 2021

Applied datasets can vary from a few hundred to thousands of samples in typical quantitative structure-activity/property (QSAR/QSPR) relationships and classification. However, the size train/test split ratios greatly affect outcome models, thus classification performance itself. We compared several combinations dataset sizes with five different machine learning algorithms find differences or similarities select best parameter settings nonbinary (multiclass) It is also known that models are ranked differently according merit(s) used. Here, 25 parameters were calculated for each model, then factorial ANOVA was applied compare results. The results clearly show not just between but lesser extent ratios. XGBoost algorithm could outperform others, even multiclass modeling. reacted change sample set size; some them much more sensitive this factor than others. Moreover, significant be detected as well, exerting great effect on test validation our models.

Language: Английский

---

Introduction to Machine Learning

Studies in computational intelligence, Journal Year: 2024, Volume and Issue: unknown, P. 51 - 94

Published: Jan. 1, 2024

Language: Английский

---

Quality prediction of seabream (SPARUS AURATA) by DEEP learning algorithms and explainable artificial intelligence

Food Chemistry, Journal Year: 2025, Volume and Issue: 474, P. 143150 - 143150

Published: Jan. 31, 2025

Language: Английский

---

Machine learning in onco-pharmacogenomics: a path to precision medicine with many challenges

Frontiers in Pharmacology, Journal Year: 2024, Volume and Issue: 14

Published: Jan. 9, 2024

Over the past two decades, Next-Generation Sequencing (NGS) has revolutionized approach to cancer research. Applications of NGS include identification tumor specific alterations that can influence pathobiology and also impact diagnosis, prognosis therapeutic options. Pharmacogenomics (PGx) studies role inheritance individual genetic patterns in drug response taken advantage technology as it provides access high-throughput data can, however, be difficult manage. Machine learning (ML) recently been used life sciences discover hidden from complex solve various PGx problems. In this review, we provide a comprehensive overview approaches employed different implicating use data. We an excursus ML algorithms exert fundamental strategies field improve personalized medicine cancer.

Language: Английский

---

TERL: classification of transposable elements by convolutional neural networks

Briefings in Bioinformatics, Journal Year: 2020, Volume and Issue: 22(3)

Published: July 22, 2020

Abstract Transposable elements (TEs) are the most represented sequences occurring in eukaryotic genomes. Few methods provide classification of these into deeper levels, such as superfamily level, which could useful and detailed information about sequences. Most that classify TE use handcrafted features k-mers homology-based search, be inefficient for classifying non-homologous Here we propose an approach, called transposable pepresentation learner (TERL), preprocesses transforms one-dimensional two-dimensional space data (i.e., image-like sequences) apply it to deep convolutional neural networks. This method tries learn best representation input correctly. We have conducted six experiments test performance TERL against other methods. Our approach obtained macro mean accuracies F1-score 96.4% 85.8% superfamilies 95.7% 91.5% order from RepBase, respectively. also 95.0% 70.6% seven databases level 89.3% 73.9% surpassed accuracy, recall specificity by on experiment with far time elapsed any all experiments. Therefore, can how predict hierarchical TEs system is 20 times three orders magnitude faster than TEclass PASTEC, respectively https://github.com/muriloHoracio/TERL. Contact:[email protected]

Language: Английский

---

Machine learning approaches to medication adherence amongst NCD patients: A systematic literature review

Informatics in Medicine Unlocked, Journal Year: 2023, Volume and Issue: 38, P. 101210 - 101210

Published: Jan. 1, 2023

Non-adherence to prescribed medication is a major public health concern that escalates the risk of morbidity and death as well incurring extra expenses associated with hospitalisation. According World Health Organization (WHO), only 50% people suffering from chronic diseases follow treatment recommendations despite counsel provided patients on importance adherence (MA). Early detection non-communicable disease (NCD) poorly adhering recommended medications using analytics based machine learning (ML) may improve outcomes NCD positively. This paper presents systematic review literature involving application ML in evaluating MA amongst patients. The articles considered this study were extracted Web Science, Google Scholar, PubMed, IEEE Explore. Twenty-five total met criteria for inclusion. These utilised techniques analyse NCDs, diabetes (n = 8), hypertension 3), cardiovascular (CVD) statin 6), cancer respiratory 2), other conditions 3). proportion days covered (PDC) was typically used evaluate MA. It emerged be high, threshold should at least 75% PDC, universally accepted threshold. In research practice, PDC ≥80% regarded high level prescription medication. Logistic regression (LR) 12), random forest (RF) 11), support vector (SVM) 7), neural net ensemble MLPs 4), XGBoost Bayesian network (BN) gradient boosting 3) most frequently applied underscored leveraging standard ML, deep (DL), has enormous potential measuring various such prediction, regression, classification, clustering. Moreover, further could conducted comprehend how alternative ML-based can measure among infectious diseases.

Language: Английский

---

Framework for Addressing Imbalanced Data in Aviation with Federated Learning

Information, Journal Year: 2025, Volume and Issue: 16(2), P. 147 - 147

Published: Feb. 16, 2025

The aviation industry generates vast amounts of data across multiple stakeholders, but critical faults and anomalies occur rarely, creating inherently imbalanced datasets that complicate machine learning applications. Traditional centralized approaches are further constrained by privacy concerns regulatory requirements limit sharing among stakeholders. This paper presents a novel framework for addressing challenges in through federated learning, focusing on fault detection, predictive maintenance, safety management. proposed combines specialized techniques handling with privacy-preserving to enable effective collaboration while maintaining security. incorporates local resampling methods, cost-sensitive weighted aggregation mechanisms improve minority class detection performance. is validated extensive experiments involving demonstrating 23% improvement accuracy 17% reduction remaining useful life prediction error compared conventional models. Results show the enhanced rare faults, improved maintenance scheduling accuracy, risk assessment distributed datasets. provides scalable practical solution using both imbalance concerns, contributing operational efficiency industry.

Language: Английский

---

Federated learning with differential privacy for breast cancer diagnosis enabling secure data sharing and model integrity

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 16, 2025

Abstract In the digital age, privacy preservation is of paramount importance while processing health-related sensitive information. This paper explores integration Federated Learning (FL) and Differential Privacy (DP) for breast cancer detection, leveraging FL’s decentralized architecture to enable collaborative model training across healthcare organizations without exposing raw patient data. To enhance privacy, DP injects statistical noise into updates made by model. mitigates adversarial attacks prevents data leakage. The proposed work uses Breast Cancer Wisconsin Diagnostic dataset address critical challenges such as heterogeneity, privacy-accuracy trade-offs, computational overhead. From experimental results, FL combined with achieves 96.1% accuracy a budget ε = 1.9, ensuring strong minimal performance trade-offs. comparison, traditional non-FL achieved 96.0% accuracy, but at cost requiring centralized storage, which poses significant risks. These findings validate feasibility privacy-preserving artificial intelligence models in real-world clinical applications, effectively balancing protection reliable medical predictions.

Language: Английский

---

Inpactor2: a software based on deep learning to identify and classify LTR-retrotransposons in plant genomes

Briefings in Bioinformatics, Journal Year: 2022, Volume and Issue: 24(1)

Published: Nov. 3, 2022

Abstract LTR-retrotransposons are the most abundant repeat sequences in plant genomes and play an important role evolution biodiversity. Their characterization is of great importance to understand their dynamics. However, identification classification these elements remains a challenge today. Moreover, current software can be relatively slow (from hours days), sometimes involve lot manual work do not reach satisfactory levels terms precision sensitivity. Here we present Inpactor2, accurate fast application that creates LTR-retrotransposon reference libraries very short time. Inpactor2 takes assembled genome as input follows hybrid approach (deep learning structure-based) detect elements, filter partial finally classify intact into superfamilies and, few tools do, lineages. This tool advantage multi-core GPU architectures decrease execution times. Using rice genome, showed run time 5 minutes (faster than other tools) has best accuracy F1-Score tested here, also having second specificity only surpassed by EDTA, but achieving 28% higher For large genomes, up seven times faster available bioinformatics tools.

Language: Английский

---

Assessment of the seismic vulnerability in an urban area with the integration of machine learning methods and GIS

Natural Hazards, Journal Year: 2025, Volume and Issue: unknown

Published: March 4, 2025

Language: Английский

---