🤖 AI Summary
This work addresses the performance degradation of atrial fibrillation (AF) classification from photoplethysmography (PPG) signals collected by wearable devices under cross-domain scenarios, where distribution shifts between source and target domains adversely affect model generalization. To tackle this challenge, the authors propose a novel framework that integrates deep generative domain adaptation with decision-theoretic uncertainty quantification. Specifically, a generative model aligns target-domain features to the source domain, while— for the first time—a task-oriented uncertainty estimation mechanism is embedded within the domain adaptation pipeline to assess the reliability of generated signals for downstream AF classification. Experimental results demonstrate that the proposed approach not only significantly improves cross-domain classification accuracy but also enhances the trustworthiness of synthesized data and the overall robustness of the system.
📝 Abstract
In principle, deep generative models can be used to perform domain adaptation; i.e. align the input feature representations of test data with that of a separate discriminative model's training data. This can help improve the discriminative model's performance on the test data. However, generative models are prone to producing hallucinations and artefacts that may degrade the quality of generated data, and therefore, predictive performance when processed by the discriminative model. While uncertainty quantification can provide a means to assess the quality of adapted data, the standard framework for evaluating the quality of predicted uncertainties may not easily extend to generative models due to the common lack of ground truths (among other reasons). Even with ground truths, this evaluation is agnostic to how the generated outputs are used on the downstream task, limiting the extent to which the uncertainty reliability analysis provides insights about the utility of the uncertainties with respect to the intended use case of the adapted examples. Here, we describe how decision-theoretic uncertainty quantification can address these concerns and provide a convenient framework for evaluating the trustworthiness of generated outputs, in particular, for domain adaptation. We consider a case study in photoplethysmography time series denoising for Atrial Fibrillation classification. This formalises a well-known heuristic method of using a downstream classifier to assess the quality of generated outputs.