Domain Adaptation for Time-Series Classification to Mitigate Covariate Shift
This addresses domain adaptation for time-series data, which is important for applications like handwriting recognition, but appears incremental as it builds on existing optimal transport and embedding techniques.
The paper tackles domain shift in time-series classification by proposing a supervised domain adaptation method that finds optimal class-dependent transformations using optimal transport techniques and selects transformations at inference using embedding similarity. The method was evaluated on simulated and online handwriting datasets, showing performance improvements.
The performance of a machine learning model degrades when it is applied to data from a similar but different domain than the data it has initially been trained on. To mitigate this domain shift problem, domain adaptation (DA) techniques search for an optimal transformation that converts the (current) input data from a source domain to a target domain to learn a domain-invariant representation that reduces domain discrepancy. This paper proposes a novel supervised DA based on two steps. First, we search for an optimal class-dependent transformation from the source to the target domain from a few samples. We consider optimal transport methods such as the earth mover's distance, Sinkhorn transport and correlation alignment. Second, we use embedding similarity techniques to select the corresponding transformation at inference. We use correlation metrics and higher-order moment matching techniques. We conduct an extensive evaluation on time-series datasets with domain shift including simulated and various online handwriting datasets to demonstrate the performance.