A plithogenic-fuzzy multilayer perceptron (PN- MLP) integrating fuzzy, intuitionistic-fuzzy, neutrosophic and plithogenic logics with XAI 2.0 for a breast cancer CAD platform

Yu. V. Trofimov, I. P. Muravyov, A. N. Averkin, A. D. Lebedev, E. M. Kuznetsov, I. A. Trusov, A. K. Alekseev, A. V. Shevchenko

Dubna State University, Dubna

Abstract: Fine-needle aspiration biopsy remains pivotal for the early detection of breast cancer. This study introduces a multiparadigm diagnostic architecture that integrates four complementary set-theoretic formalisms-fuzzy, intuitionistic fuzzy, neutrosophic, and plithogenic sets-with a plithogenic-fuzzy multilayer perceptron (PN-MLP) and an embedded XAI 2.0 explanation layer. The proposed workflow encompasses: (i) multiformal fuzzification of morphometric features extracted from an augmented Breast Cancer Wisconsin corpus; (ii) weighted aggregation using plithogenic operators that adaptively attenuate inter-attribute conflict; and (iii) an ensemble classification stage that preserves the full semantic vector of certainty, contradiction, and indeterminacy. End-to-end incorporation of plithogenic reasoning markedly improves epistemic reliability while maintaining clinician-oriented interpretability, thereby establishing a benchmark prototype for trustworthy next-generation explainable artificial intelligence in oncological cytology.

Keywords: breast cancer, fuzzy logic, intuitionistic fuzzy sets, neutrosophic sets, plithogenic sets, LIME, SHAP, XAI, machine learning, neural networks, perceptron, fuzzy perceptron.

UDC: 004.8

Received: 12.05.2025
Revised: 04.06.2025

DOI: 10.26456/fssc132

Bibliographic databases:

•