Minimum Displacement in Existing Moment (MDEM)- A new supervised learning algorithm by incrementally constructing the moments of the underlying classes

We introduce a supervised learning method that classifies each test point by selecting the class for which its inclusion causes minimum displacement of the class's existing n-th central moment. After each such inclusion, the n-th central moment of the corresponding class is updated by some incr...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:PloS one Ročník 20; číslo 12; s. e0336933
Hlavní autor: Nizam, Ahmed Mehedi
Médium: Journal Article
Jazyk:angličtina
Vydáno: United States Public Library of Science 01.12.2025
Témata:
Algorithms
Breast cancer
Breast Neoplasms
Classification
Data points
Datasets
Decision trees
Diabetes Mellitus
Expected values
Female
Humans
Logistic Models
Machine learning
Neural networks
Neural Networks, Computer
Optimization techniques
Random variables
Regression analysis
Sample size
Supervised learning
Supervised Machine Learning
Support Vector Machine
Support vector machines
ISSN:1932-6203
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:We introduce a supervised learning method that classifies each test point by selecting the class for which its inclusion causes minimum displacement of the class's existing n-th central moment. After each such inclusion, the n-th central moment of the corresponding class is updated by some incremental calculations in constant time, i.e., each class evolves gradually and changes its definition incrementally after the inclusion of every new data point. We then use k-fold and stratified k-fold cross validation techniques to compare the performance of our proposed model with various state of the art supervised learning algorithms including Neural Network (NN), Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbor (KNN) and Logistic Regression (LR) using Pima Indian Diabetes (PID) dataset and Wisconsin breast cancer dataset, which are popular datasets in machine learning research. Our analyses suggest the performances of different MDEM algorithms as proposed here involving different order of moments vary within the range of [83.19% - 95.82%] of the best algorithm under consideration in k-fold and stratified k-fold cross validation techniques for PID dataset. Moreover, for Wisconsin breast cancer dataset, different variants of MDEM algorithms have achieved accuracy scores in the range of [88.85% - 96.41%] of the best algorithm. Finally, we compare the results produced by different algorithms by constructing the corresponding confusion matrices.
ISSN:1932-6203
DOI:10.1371/journal.pone.0336933