0 likes | 2 Views
COMPUTER REPAIR & SUPPORT<br>Table of contents<br>Introduction<br>XAI<br>History of XAI<br>Difference Between AI and<br>XAICOMPUTER REPAIR & SUPPORT<br>Table of contents<br>XAI Grown<br>Importance of XAI<br>Applications of XAI<br>XAI Tools and TechniquesIntroduction<br>Explainable AI (XAI) refers to a set of processes and<br>methods that allow human users to comprehend and<br>trust the results and output created by machine learning<br>algorithms. XAI seeks to make the decision-making<br>process of AI models transparent, interpretable, and<br>understandable.XAI<br>Stands for: Explainable Artificial Intelligence<br>Focuses on making AI models understandable to<br>humans<br>Answers questions like:<br>Why did the model make this prediction?<br>Which features were most influential?<br>How confident is the model in its output?History of XAI<br>1980s: Emerged in expert systems; explanation<br>was simpler due to rule-based logic.<br>2010s: Deep learning models became<br>widespread, causing concern due to their blackbox nature.<br>2016: DARPA launched the XAI program to<br>promote transparency in AI systems.<br>Today: XAI is a core part of Responsible AI and<br>ethical machine learning.Difference Between AI and XAI<br>Aspect AI XAI<br>Goal Optimize accuracy Improve transparency<br>and interpretability<br>Model type Often black-box Post-hoc explanation or<br>inherently interpretable<br>Audience Machines, developers Humans, regulators,<br>domain experts<br>Key Question What is the output? Why is this the output?XAI Grown<br>1. Trust: Users need to understand decisions to trust AI.<br>2. Legal Requirements: GDPR and similar laws require<br>explainability.<br>3. Debugging and Optimization: Helps developers identify<br>flaws.<br>4. Fairness and Bias Detection: Prevents discrimination in<br>sensitive domains.Importance of XAI<br>u2022 Enables ethical and responsible AI<br>u2022 Ensures accountability and transparency<br>u2022 Critical in high-stakes fields:<br>- Healthcare<br>- Finance<br>- Criminal Justice<br>- Autonomous SystemsApplications of XAI<br>Sector Application Example<br>Healthcare Explaining disease risk<br>predictions<br>Finance Understanding loan<br>approval/rejection<br>Automotive Explaining autonomous vehicle<br>actions<br>Security Justifying automated surveillance<br>alerts<br>Recommendations Clarifying why a product or movie<br>was suggestedXAI Tools and Techniques<br>Post-hoc Explanation Tools:<br>- SHAP: Uses game theory to assign feature importance<br>- LIME: Creates simple, local surrogate models for<br>explanation<br>- Grad-CAM: Highlights important regions in images<br>(vision models)<br>Interpretable Models:<br>- Decision Trees<br>- Logistic Regression<br>- Rule-based Systems
E N D