Combining Statistical Features and Deep Encodings for Rehearsal-Based Class-Incremental Time Series Classification
For practitioners needing to incrementally add new time series classes without forgetting old ones, this work offers a rehearsal-based approach that balances accuracy and forgetting, though it is incremental in nature.
The paper tackles class-incremental continual learning for multivariate time series classification, proposing a dual-stream feature extraction pipeline combining deep temporal embeddings from a frozen foundation model with statistical features. The method achieves competitive average accuracy and low forgetting rates across five benchmark datasets.
Many systems used in real-world environments require adding new categories and incorporating new information without forgetting what was previously learnt by the classification model. This is known as class-incremental continual learning, and in the case of multivariate time-series, is further complicated by the temporal structure of the data. In this paper, we present a novel approach for performing class incremental continual learning for the classification of multivariate time series data based upon the construction of a dual-stream feature extraction pipeline (using both deep temporal embedding features generated via a pre-trained frozen foundation model and application of statistical features). Evaluated on five benchmark datasets, the proposed system achieves competitive average accuracy across all datasets while maintaining low forgetting rates across all experimental configurations.