Predicting Survival Outcomes in the Presence of Unlabeled Data

Many clinical studies require the follow-up of patients over time. This is challenging: apart from frequently observed drop-out, there are often also organizational and financial challenges, which can lead to reduced data collection and, in turn, can complicate subsequent analyses. In contrast, there is often plenty of baseline data available of patients with similar characteristics and background information, e.g., from patients that fall outside the study time window. In this article, we investigate whether we can benefit from the inclusion of such unlabeled data instances to predict accurate survival times. In other words, we introduce a third level of supervision in the context of survival analysis, apart from fully observed and censored instances, we also include unlabeled instances. We propose three approaches to deal with this novel setting and provide an empirical comparison over fifteen real-life clinical and gene expression survival datasets. Our results demonstrate that all approaches are able to increase the predictive performance over independent test data. We also show that integrating the partial supervision provided by censored data in a semi-supervised wrapper approach generally provides the best results, often achieving high improvements, compared to not using unlabeled data.

     

Decision trees for hierarchical multi-label classification

Hierarchical multi-label classification (HMC) is a variant of classification where instances may belong to multiple classes at the same time and these classes are organized in a hierarchy. This article presents several approaches to the induction of decision trees for HMC, as well as an empirical study of their use in functional genomics. We compare learning a single HMC tree (which makes predictions for all classes together) to two approaches that learn a set of regular classification trees (one for each class). The first approach defines an independent single-label classification task for each class (SC). Obviously, the hierarchy introduces dependencies between the classes. While they are ignored by the first approach, they are exploited by the second approach, named hierarchical single-label classification (HSC). Depending on the application at hand, the hierarchy of classes can be such that each class has at most one parent (tree structure) or such that classes may have multiple parents (DAG structure). The latter case has not been considered before and we show how the HMC and HSC approaches can be modified to support this setting. We compare the three approaches on 24 yeast data sets using as classification schemes MIPS’s FunCat (tree structure) and the Gene Ontology (DAG structure). We show that HMC trees outperform HSC and SC trees along three dimensions: predictive accuracy, model size, and induction time. We conclude that HMC trees should definitely be considered in HMC tasks where interpretable models are desired.     

Network representation with clustering tree features

This paper studies the relationship between a case-based decision theory (CBDT) and an ideal point model (IPM). We show that a case-based decision model (CBDM) can be transformed into an IPM under some assumptions. This transformation can allow us to visualize the relationship among data and simplify the calculations of distance between one current datum and the ideal point, rather than the distances between data. Our results will assist researchers with their product design analysis and positioning of goods through CBDT, by revealing past dependences or providing a reference point. Furthermore, to check whether the similarity function, presented in the theoretical part, is valid for empirical analysis, we use data on the viewing behavior of audiences of TV dramas in Japan and compare the estimation results under the CBDM that corresponds to a standard decision model with similarities and other various similarity functions and without a similarity function. Our empirical analysis shows that the CBDM with a similarity function, presented in this study, best fits the data.     

First order random forests: Learning relational classifiers with complex aggregates

In relational learning, predictions for an individual are based not only on its own properties but also on the properties of a set of related individuals. Relational classifiers differ with respect to how they handle these sets: some use properties of the set as a whole (using aggregation), some refer to properties of specific individuals of the set, however, most classifiers do not combine both. This imposes an undesirable bias on these learners. This article describes a learning approach that avoids this bias, using first order random forests. Essentially, an ensemble of decision trees is constructed in which tests are first order logic queries. These queries may contain aggregate functions, the argument of which may again be a first order logic query. The introduction of aggregate functions in first order logic, as well as upgrading the forest’s uniform feature sampling procedure to the space of first order logic, generates a number of complications. We address these and propose a solution for them. The resulting first order random forest induction algorithm has been implemented and integrated in the ACE-ilProlog system, and experimentally evaluated on a variety of datasets. The results indicate that first order random forests with complex aggregates are an efficient and effective approach towards learning relational classifiers that involve aggregates over complex selections. Editor: Rui Camacho     

Outlier detection in relational data: A case study in geographical information systems

Geographical information systems are commonly used for a variety of purposes. Many of them make use of a large database of geographical data, the correctness of which strongly influences the reliability of the system. In this paper, we present an approach to quality maintenance that is based on automatic discovery of non-perfect regularities in the data. The underlying idea is that exceptions to these regularities (‘outliers’) are considered probable errors in the data, to be investigated by a human expert. A case study shows how the tool can be used for extracting valuable knowledge about outliers in real-world geographical data, in an adaptive manner to the evolving data model supporting it. While the tool aims specifically at geographical information systems, the underlying approach is more broadly applicable for quality maintenance in data-rich intelligent systems.