kanomaly

https://medium.com/@ronantech/time-series-anomaly-detection-using-kolmogorov-arnold-networks-abd9bbeaa9af

Time Series Anomaly Detection using a Kolmogorov-Arnold Network

Results:

Synthetic Data:

Test Loss: 0.0282, Test Acc: 0.9100, Precision: 1.0000, Recall: 0.5714, F1: 0.7273, ROC AUC: 0.8758

ECG5000 Time Series Anomaly Dataset:

Test Loss: 0.0627, Test Acc: 0.8270, Precision: 0.7280, Recall: 0.9327, F1: 0.8177, ROC AUC: 0.8890