Noob here, learning, excited to be part of KAN.

[cbroker1]

In your readme, you mention that because KAN is a fully connected network, it will not be able to detect temporal dependencies. Forgive me, as I just had to look up what temporal dependencies meant, but what would it take for such to be possible with KAN? Bravo on the repo, hope to be at your level someday.

[hadifawaz1999]

Hello , thank you for taking interest in our work !
For time series data, it has been seen that its better to use architectures that can extract local temporal dependencies and feature from the samples [1].
Until now the best architecture to be used for time series classification is the convolution based one.
For this reason we believe that if KAN is extended in the future for a KANConv layer, than it would be interesting to compare it to normal convolution layers that we currently use

[1] Ismail Fawaz, Hassan, et al. "Deep learning for time series classification: a review." Data mining and knowledge discovery 33.4 (2019): 917-963.