Article
Mathematics
Sebastian Alberto Grillo, Jose Luis Vazquez Noguera, Julio Cesar Mello Roman, Miguel Garcia-Torres, Jacques Facon, Diego P. Pinto-Roa, Luis Salgueiro Romero, Francisco Gomez-Vela, Laura Raquel Bareiro Paniagua, Deysi Natalia Leguizamon Correa
Summary: This study analyzes the impact of redundant features on classification model performance and proposes a theoretical framework for analyzing feature construction and selection. The experimental results suggest that a large number of redundant features can reduce the classification error.
Article
Biochemical Research Methods
Fengsheng Wang, Leyi Wei
Summary: In this study, we propose a novel multi-scale end-to-end deep learning model, MSTLoc, for identifying protein subcellular locations in the imbalanced multi-label immunohistochemistry (IHC) images dataset. We demonstrate that the proposed MSTLoc outperforms current state-of-the-art models in multi-label subcellular location prediction. Through feature visualization and interpretation analysis, we show that the multi-scale deep features learned from our model exhibit better ability in capturing discriminative patterns underlying protein subcellular locations, and the features from different scales are complementary for the improvement in performance. Case study results indicate that our MSTLoc can successfully identify some biomarkers from proteins that are closely involved in cancer development.
Article
Mathematics
Adel Fahad Alrasheedi, Khalid Abdulaziz Alnowibet, Akash Saxena, Karam M. Sallam, Ali Wagdy Mohamed
Summary: In this study, a chaos embed marine predator algorithm (CMPA) is proposed for feature selection in data mining applications. The comparative analysis and statistical significance tests provide evidence for the effectiveness and applicability of the proposed algorithm.
Article
Computer Science, Artificial Intelligence
Zhaogeng Liu, Jielong Yang, Li Wang, Yi Chang
Summary: In this paper, a novel wrapper feature selection method named ERASE is proposed, which learns and utilizes sample relations and feature relations for feature selection. Experimental results demonstrate that our method outperforms other feature selection methods in most cases.
PATTERN RECOGNITION
(2023)
Article
Automation & Control Systems
Xiaokang Wang, Huiwen Wang, Dexiang Wu
Summary: This study proposes an online dynamic feature weighting algorithm to monitor feature drift in data streams. The algorithm detects changes in class relevance of features based on the log-likelihood divergence score, and it has been shown to improve the accuracy rates of Nearest Neighbor and Naive Bayes classifiers on both synthetic and real-world datasets.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2022)
Article
Computer Science, Artificial Intelligence
Lianxi Wang, Shengyi Jiang, Siyu Jiang
Summary: The study introduces a novel feature selection algorithm that selects relevant and interactive features using a maximum criterion, leading to improved classification accuracy. Experimental results show that the algorithm efficiently selects features and enhances classifiers to achieve better or comparable classification accuracy compared to ten representative competing feature selection algorithms.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Clinical Neurology
E. I. S. Hofmeijer, C. O. Tan, F. van der Heijden, R. Gupta
Summary: Researchers tested ensemble learning for selecting the best artificial intelligence models for intracranial hemorrhage detection, but ensemble learning methods did not outperform the single best model.
AMERICAN JOURNAL OF NEURORADIOLOGY
(2023)
Article
Computer Science, Information Systems
G. Wiselin Jiji
Summary: Algorithms in computer vision are crucial for extracting valuable hidden information from datasets. This study focuses on diagnosing neurodegenerative diseases such as Alzheimer's disease, Parkinson's disease, and bipolar disorder. It uses potential biomarkers extracted from T1 MRI and brain tissue volumes, specifically the 3D Speeded Up Robust Feature (SURF) and 3D Scale Invariant Feature Transform (SIFT) features. Random Forest and SVM approaches are employed to select key points for diagnosis, achieving a classification accuracy of 98.6%.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Gazal, Kapil Juneja
Summary: This paper investigates a two-level filter-based hybrid model to accurately identify spam messages. The model selects the most important features through filtering and evaluation methods, and uses classifiers to generate probabilistic scores for spam detection. The experimental results show that the model achieves high accuracy on multiple datasets and outperforms traditional methods.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2022)
Article
Computer Science, Information Systems
Amina Benkessirat, Nadjia Benblidia
Summary: In real-life classification applications, it can be challenging to select model features that adequately classify samples from a large number of candidates. This article's main contributions include evaluating the relevance and redundancy of features, defining the feature selection problem as an eigenvalue computation problem with a linear constraint, and efficiently selecting the best features. The approach was tested on 20 UCI benchmark datasets and compared with other widely used and state-of-the-art approaches. The experimental results showed that our approach improved the classification task by using only 20% of the conventional features.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Xi-Ao Ma, Hao Xu, Chunhua Ju
Summary: This paper proposes a class-specific feature selection method based on information theory. A class-specific feature evaluation criterion called CSMDCCMR is developed, and a feature selection algorithm is designed to select a suitable feature subset for each class. Experimental results demonstrate the superiority of the proposed method compared to other methods.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Chemistry, Physical
Xiang Huang, Shengluo Ma, C. Y. Zhao, Hong Wang, Shenghong Ju
Summary: This study proposes a high-throughput screening framework for designing polymer chains with high thermal conductivity using interpretable machine learning and physical feature engineering. By optimizing physical descriptors and assisting machine learning models, the framework achieves higher prediction accuracy compared to traditional methods. The study also analyzes the contributions of individual descriptors and derives an explicit prediction equation for thermal conductivity. Polymer chains with high thermal conductivity are predominantly pi-conjugated structures with strong intra-chain interactions, resulting in enhanced thermal transport.
NPJ COMPUTATIONAL MATERIALS
(2023)
Article
Computer Science, Artificial Intelligence
Mostafa Khojastehnazhand, Mozaffar Roostaei
Summary: This study used a machine vision system and texture feature extraction methods to classify seven varieties of wheat in the East Azerbaijan Province of Iran. By utilizing unsupervised and supervised methods, along with feature extraction, the different wheat varieties were identified with over 95% accuracy.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Computer Science, Information Systems
Cun Ji, Mingsen Du, Yanxuan Wei, Yupeng Hu, Shijun Liu, Li Pan, Xiangwei Zheng
Summary: Time series classification is widely used in various domains, including EEG/ECG classification, device anomaly detection, and speaker authentication. Despite the existence of many methods, selecting intuitive temporal features for accurate classification remains a challenge. Therefore, this paper proposes a new method called TSC-RTF, which utilizes random temporal features, and shows that it can compete with state-of-the-art methods.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2023)
Article
Engineering, Multidisciplinary
M. Shaheen, N. Naheed, A. Ahsan
Summary: Big data analytics uncovers hidden patterns through classification, prediction and reinforcement of big datasets. Relevant, important and informative features are selected using different filtration techniques. A new feature selection technique called Relevance-diversity algorithm and a new supervised classification algorithm based on Naive Bayes classification are proposed. The performance of these techniques is evaluated using various datasets, and the results show improvements in terms of feature selection, accuracy, and time complexity.
ALEXANDRIA ENGINEERING JOURNAL
(2023)