Summary
Explaining the logic behind a machine learning model’s decision is crucial to increasing transparency and interpretability to earn users’ trust and meet regulatory compliance. In contrast, demonstrating failures in an AI model is equally important to improve model performance and avoid detrimental impacts in high-stakes settings.
Finding anomalies can be tedious, since there are various anomalies in time series data, such as seasonal and quantile anomalies. In this chapter, you learned about detecting and explaining anomalies for time series data with deep learning. I hope you can apply what you have learned about time series anomaly explainability to identify actionable insights.
The next chapter will cover computer vision anomaly explainability.