Multi-Resolution Weak Supervision for Sequential Data
This addresses a key limitation in machine learning for domains like biomedical video analysis, where expert-labeled data is scarce, by enabling effective weak supervision for sequential data, though it is incremental as it builds on existing weak supervision approaches.
The paper tackles the problem of labeling sequential data like video using multi-resolution weak supervision sources, which existing methods fail to model, by proposing Dugong, a framework that uniquely recovers accuracy and correlation parameters and improves sample complexity. It results in a 36.8 F1 point improvement over traditional supervision in biomedical video labeling and averages 16.0 and 24.2 F1 point gains over traditional and existing weak supervision methods across tasks.
Since manually labeling training data is slow and expensive, recent industrial and scientific research efforts have turned to weaker or noisier forms of supervision sources. However, existing weak supervision approaches fail to model multi-resolution sources for sequential data, like video, that can assign labels to individual elements or collections of elements in a sequence. A key challenge in weak supervision is estimating the unknown accuracies and correlations of these sources without using labeled data. Multi-resolution sources exacerbate this challenge due to complex correlations and sample complexity that scales in the length of the sequence. We propose Dugong, the first framework to model multi-resolution weak supervision sources with complex correlations to assign probabilistic labels to training data. Theoretically, we prove that Dugong, under mild conditions, can uniquely recover the unobserved accuracy and correlation parameters and use parameter sharing to improve sample complexity. Our method assigns clinician-validated labels to population-scale biomedical video repositories, helping outperform traditional supervision by 36.8 F1 points and addressing a key use case where machine learning has been severely limited by the lack of expert labeled data. On average, Dugong improves over traditional supervision by 16.0 F1 points and existing weak supervision approaches by 24.2 F1 points across several video and sensor classification tasks.