Time series classification with random convolution kernels: pooling operators and input representations matter
This work addresses the problem of efficient and accurate time series classification for researchers and practitioners, though it is incremental as it builds on the MiniRocket framework.
The authors tackled time series classification by introducing SelF-Rocket, a method that dynamically selects optimal input representations and pooling operators during training, achieving state-of-the-art accuracy on the UCR benchmark datasets.
This article presents a new approach based on MiniRocket, called SelF-Rocket, for fast time series classification (TSC). Unlike existing approaches based on random convolution kernels, it dynamically selects the best couple of input representations and pooling operator during the training process. SelF-Rocket achieves state-of-the-art accuracy on the University of California Riverside (UCR) TSC benchmark datasets.