Abstract
Online structure learning approaches, such as those stemming from statistical relational learning, enable the discovery of complex relations in noisy data streams. However, these methods assume the existence of fully-labelled training data, which is unrealistic for most real-world applications. We present a novel approach for completing the supervision of a semi-supervised structure learning task. We incorporate graph-cut minimisation, a technique that derives labels for unlabelled data, based on their distance to their labelled counterparts. In order to adapt graph-cut minimisation to first order logic, we employ a suitable structural distance for measuring the distance between sets of logical atoms. The labelling process is achieved online (single-pass) by means of a caching mechanism and the Hoeffding bound, a statistical tool to approximate globally-optimal decisions from locally-optimal ones. We evaluate our approach on the task of composite event recognition by using a benchmark dataset for human activity recognition, as well as a real dataset for maritime monitoring. The evaluation suggests that our approach can effectively complete the missing labels and eventually, improve the accuracy of the underlying structure learning system.
Highlights
Methods for handling both uncertainty and complex relational structure have received much attention in machine learning
To demonstrate the benefits of SPLICE, our proposed method to semisupervised online structure learning, we focus on composite event recognition (CER), by employing the OSLα (Michelioudakis et al 2016b) and OLED (Katzouris et al 2016) online structure learners
The 100% setting in the bottom figures corresponds to full supervision, i.e., no unlabelled instances to be completed by SPLICE
Summary
Methods for handling both uncertainty and complex relational structure have received much attention in machine learning. Online structure learning approaches for MLNs have been effectively applied to a variety of tasks (Michelioudakis et al 2016b; Huynh and Mooney 2011) These approaches facilitate the automated discovery of multi-relational dependencies in noisy environments, they assume a fully labelled training sequence, which is unrealistic in most real-world applications. DEC includes the core domain-independent axioms of the Event Calculus, which determine whether a fluent holds or not at a specific timepoint. This axiomatisation incorporates the common sense law of inertia, according to which fluents persist over time, unless they are affected by an event occurrence. The core DEC axioms are defined as follows: HoldsAt( f , t+1) ⇐ InitiatedAt( f , t)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
