Demystifying AI: A Robust and Comprehensive Approach to Explainable AI

Abstract

The adoption of Artificial Intelligence (AI) and Machine Learning (ML) in various computing platforms and areas, necessitates the development of strong Explainable AI (XAI) techniques. Most current AI models are opaque about their decision-making process thereby impeding trust, debugging, and improvement. The goal of this research is to develop comprehensive robust XAI methods capable of explaining the reasoning and decision-making processes in Autonomic, Edge, Server-less, Quantum computing platforms and IoT, Business Automation, Service Innovation domains where these AI models are deployed.This study comprehensively addresses the opacity in AI models through solutions for balanced test-train splits, model evaluation, feature importance, metric imbalances, ROC curve and precision-recall curve analysis, accuracy and statistical metrics, benefits of manual review. This research aims at increasing transparency and trustworthiness within AI systems through developing as well as applying such XAI methods that can detect and mitigate biases while enhancing ethical debugging; responsible development for AI enabled computing purposes.

Ähnliche Arbeiten