HDC-MiniROCKET: Explicit Time Encoding in Time Series Classification with Hyperdimensional Computing
This work addresses time series classification for domains requiring accurate and efficient methods, but it is incremental as it builds upon an existing approach.
The authors tackled the problem of time series classification by extending MiniROCKET with hyperdimensional computing to improve global temporal encodings, achieving better results on datasets with high temporal dependence without increasing computational effort for inference, as demonstrated on 128 UCR benchmark datasets.
Classification of time series data is an important task for many application domains. One of the best existing methods for this task, in terms of accuracy and computation time, is MiniROCKET. In this work, we extend this approach to provide better global temporal encodings using hyperdimensional computing (HDC) mechanisms. HDC (also known as Vector Symbolic Architectures, VSA) is a general method to explicitly represent and process information in high-dimensional vectors. It has previously been used successfully in combination with deep neural networks and other signal processing algorithms. We argue that the internal high-dimensional representation of MiniROCKET is well suited to be complemented by the algebra of HDC. This leads to a more general formulation, HDC-MiniROCKET, where the original algorithm is only a special case. We will discuss and demonstrate that HDC-MiniROCKET can systematically overcome catastrophic failures of MiniROCKET on simple synthetic datasets. These results are confirmed by experiments on the 128 datasets from the UCR time series classification benchmark. The extension with HDC can achieve considerably better results on datasets with high temporal dependence without increasing the computational effort for inference.