Machine learning-assisted optimization of food-grade spirulina cultivation in seawater-based media: From laboratory to large-scale production

Journal of Environmental Management, Год журнала: 2024, Номер 369, С. 122279 - 122279

Опубликована: Авг. 31, 2024

Язык: Английский

---

EFFECTIVENESS OF VARIABLE SELECTION METHODS FOR MACHINE LEARNING AND CLASSICAL STATISTICAL MODELS

Metody Ilościowe w Badaniach Ekonomicznych, Год журнала: 2024, Номер 25(2), С. 58 - 69

Опубликована: Июнь 18, 2024

In line with new international financial supervision directives (IFRS9), banks should look at a set of analytical tools, such as machine learning. The introduction these methods into banking practice requires reformulation business goals, both in terms the accuracy predictions and definition risk factors. article compares for selecting variables assigning "importance" statistical algorithmic models. calculations were carried out using example data classification loan default. effectiveness various learning algorithms on selected sets was compared. results analyzes indicate need to revise concept variable so that it does not depend structure model.

Язык: Английский

---

L2R-MLP: A Multilabel Classification Scheme for the Detection of DNS Tunneling

Data Science and Management, Год журнала: 2024, Номер unknown

Опубликована: Ноя. 1, 2024

Язык: Английский

---

Ensemble Algorithm Based on Gene Selection, Data Augmentation, and Boosting Approaches for Ovarian Cancer Classification

Diagnostics, Год журнала: 2024, Номер 14(24), С. 2772 - 2772

Опубликована: Дек. 10, 2024

Background: Ovarian cancer is a difficult and lethal illness that requires early detection precise classification for effective therapy. Microarray technology has permitted the simultaneous assessment of hundreds genes’ expression levels, yielding important insights into molecular pathways driving ovarian cancer. To reduce computational complexity improve accuracy, choosing most likely differential genes to explain impacts necessary. Medical datasets, including those related cancer, are often limited in size due privacy concerns, data collection challenges, rarity certain conditions. Data augmentation allows researchers expand dataset, providing larger more diverse set examples model training. Recent advances machine learning bioinformatics have shown promise improving based on gene information. Methods: In this paper, we present an ensemble algorithm selection, augmentation, boosting approaches classification. proposed approach, initial genetic were first subjected feature selection. Results: The target screened combined with algorithms. From results, chosen ten could accurately classify at 98.21%. Conclusions: We further show clustering real-world data, 100% accuracy strong performance distinguishing between distinct subtypes. may help doctors identify patients develop individualized treatment plans.

Язык: Английский

---

CNI-VIF: Enhanced Feature Selection for Graph Databases by Integrating Composite Node Information in VIF

International Journal of Electrical and Electronics Engineering, Год журнала: 2024, Номер 11(11), С. 100 - 113

Опубликована: Ноя. 30, 2024

Feature selection and dimensionality reduction are critical techniques in today's data-centric world, where vast complex datasets necessitate efficient effective methods for analysis decision-making. In this research, an enhanced feature technique, Composite Node Information - Variance Inflation Factor (CNI-VIF), tailored graph databases, which particularly focuses on network traffic datasets, is proposed. Traditional often fail to adequately capture the interrelationships data. The proposed method incorporates (CNI), aggregate of Betweenness, Closeness, Degree centrality, into VIF framework address these limitations. By integrating CNI, not only improves graph-based features but also achieves decreased computation time, making process more efficient. Experiments conducted CTU-13, IoT-23, NCC-2 demonstrate that CNI-VIF significantly outperforms traditional by effectively selecting features, thus enhancing performance machine learning models. Specifically, Random Forest algorithm shows exceptional results among all techniques, with yielding best overall. indicate offering a robust mechanism enhances model predictive accuracy.

Язык: Английский

---

Antimicrobial Peptides as Broad-Spectrum Therapeutics: Computational Analysis to Identify Universal Physical-Chemical Features Responsible for Multitarget Activity

The Journal of Physical Chemistry Letters, Год журнала: 2024, Номер 15(50), С. 12416 - 12424

Опубликована: Дек. 11, 2024

Antimicrobial peptides (AMPs) hold significant potential as broad-spectrum therapeutics due to their ability target a variety of different pathogens, including bacteria, fungi, and viruses. However, the rational design these requires molecular understanding properties that enable such activity. In this study, we present computational analysis utilizes machine-learning methods distinguish with single-target activity from those against multiple pathogens. By optimizing feature-selection procedure, most relevant physical-chemical properties, dipeptide compositions, solvent accessibility, charge distributions, optimal hydrophobicity, differentiate between narrow-spectrum are identified. Possible scenarios responsible for universality features discussed. These findings provide valuable insights into mechanisms multitarget AMPs.

Язык: Английский

---

Emotional and Sarcastic Sentiment Analytics - An Extreme AI Model

Опубликована: Сен. 11, 2024

Язык: Английский

---