Uncertainty-Aware Multiple Instance Learning from Large-Scale Long Time Series Data

TL;DR

This paper introduces an uncertainty-aware multiple instance learning framework for large-scale long time series classification, automatically identifying relevant periods and improving vessel detection accuracy through uncertainty-based attention and data fusion.

Contribution

The novel framework leverages predictive uncertainty to automatically focus on relevant time segments and fuse multiple data modalities, enhancing classification performance on large-scale long time series data.

Findings

Effective detection of vessel types from trajectory data

Uncertainty-aware fusion improves detection accuracy

Framework identifies relevant periods automatically

Abstract

We propose a novel framework to classify large-scale time series data with long duration. Long time seriesclassification (L-TSC) is a challenging problem because the dataoften contains a large amount of irrelevant information to theclassification target. The irrelevant period degrades the classifica-tion performance while the relevance is unknown to the system.This paper proposes an uncertainty-aware multiple instancelearning (MIL) framework to identify the most relevant periodautomatically. The predictive uncertainty enables designing anattention mechanism that forces the MIL model to learn from thepossibly discriminant period. Moreover, the predicted uncertaintyyields a principled estimator to identify whether a prediction istrustworthy or not. We further incorporate another modality toaccommodate unreliable predictions by training a separate modelbased on its availability and conduct uncertainty aware fusion toproduce the final prediction. Systematic evaluation is conductedon the Automatic Identification System (AIS) data, which is col-lected to identify and track real-world vessels. Empirical resultsdemonstrate that the proposed method can effectively detect thetypes of vessels based on the trajectory and the uncertainty-awarefusion with other available data modality (Synthetic-ApertureRadar or SAR imagery is used in our experiments) can furtherimprove the detection accuracy.