XAI Hybrid Local-Global Model for Brain Tumor Classification

Abstract

This repository contains the research code for an XAI Hybrid Local-Global Model for Brain Tumor Classification. The proposed approach combines Local Binary Patterns (LBP) as a local texture feature descriptor with EfficientNetB0 as a deep feature extractor to perform brain tumor classification. The repository also includes LIME (Local Interpretable Model-agnostic Explanations) examples to provide visual explanations of model predictions. In addition to the proposed hybrid EfficientNetB0-LBP model, the repository provides implementations and pretrained models for several deep learning architectures used for comparison. The repository is intended to support research reproducibility, model evaluation, and explainable AI experimentation.