Model Interpretability in MATLAB

Interpretable machine learning (or in deep learning, “Explainable AI”) provides techniques and algorithms that overcome the black-box nature of AI models. By revealing how various features contribute (or do not contribute) to predictions, you can validate that the model is using the right evidence for its predictions and reveal model biases that were not apparent during training.

Get an overview of model interpretability and the use cases it addresses. For engineers and scientists who are interested in adopting machine learning but weary of black-box models, we explain how interpretability can satisfy regulations, build trust in machine learning, and validate that models are working. That’s particularly important in industries like finance and medical devices where regulations set strict guidelines. We provide an overview of interpretability methods for machine learning and how to apply them in MATLAB®. We demonstrate interpretability in the context of a medical application, classifying heart arrhythmia based on ECG signals.

Additional Resources

Learn more about model interpretability on our discovery page: https://bit.ly/328gFHV Try out applying LIME in this example: https://bit.ly/2E9xfPA Apply Partial Dependence plots (PDP) and Individual Conditional Expectation (ICE) plots to Regression: https://bit.ly/3l42Anx