Challenges Towards Production-Ready Explainable Machine Learning
summary
Fintech deployment uses a preexisting random forest and needs both global reports for regulators and local justifications for clients and analysts, so the paper documents using model-agnostic tools like LIME and surrogate models to attach explanations without changing models. It finds most explainers target CV/NLP, fail on tabular messy datasets, and that explanations enable a data reconstruction attack, so recommendations include systematic user validation, adversarial simulation, and choosing local vs global explanation up front.
AI-written summary · about
chapters (8)
- 0:00Start
- 1:26Motivations behind Explainability and How To Use It in Machine Learning
- 4:52Explainability
- 9:10Model Dependence
- 10:11Text-Based Framework
- 11:03Challenges and Recommendation
- 12:59Model and Task Dependency
- 13:50Task Dependency
related talks
16mBuilding Trust through Explainable AI
29mTowards Explainable Machine Learning
01h 29mNavigating ML Explainability: From Theory to Implementation
46mExplaining a Machine Learning blackbox
35mWhy Do You Need an Explainable Ai?
24mWhy and How to Build Explainability into your ML Workflow
28mManaging AI Risk: Beyond Explainability
37mExplaining model explainability