CLOPS: Continual Learning of Physiological Signals

Abstract

Deep learning algorithms are known to experience destructive interference when instances violate the assumption of being independent and identically distributed (i.i.d). This violation, however, is ubiquitous in clinical settings where data are streamed temporally and from a multitude of physiological sensors. To overcome this obstacle, we propose CLOPS, a replay-based continual learning strategy. In three continual learning scenarios based on three publically-available datasets, we show that CLOPS can outperform the state-of-the-art methods, GEM and MIR. Moreover, we propose end-to-end trainable parameters, which we term task-instance parameters, that can be used to quantify task difficulty and similarity. This quantification yields insights into both network interpretability and clinical applications, where task difficulty is poorly quantified.

References in corpus (11)

• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Efficient Lifelong Learning with A-GEM
• Three scenarios for continual learning
• Bayesian Active Learning for Classification and Preference Learning
• BatchBALD: Efficient and Diverse Batch Acquisition for Deep Bayesian Active Learning
• Towards Robust Evaluations of Continual Learning
• O-MedAL: Online Active Deep Learning for Medical Image Analysis
• Toward Understanding Catastrophic Forgetting in Continual Learning
• Continual Learning for Domain Adaptation in Chest X-ray Classification
• Icebreaker: Element-wise Active Information Acquisition with Bayesian Deep Latent Gaussian Model
• Approval policies for modifications to Machine Learning-Based Software as a Medical Device: A study of bio-creep