Introduction to the Special Issue on Meta-Learning

Recent advances in meta-learning are providing the foundations to construct meta-learning assistants and task-adaptive learners. The goal of this special issue is to foster an interest in meta-learning by compiling representative work in the field. The contributions to this special issue provide strong insights into the construction of future meta-learning tools. In this introduction we present a common frame of reference to address work in meta-learning through the concept of meta-knowledge. We show how meta-learning can be simply defined as the process of exploiting knowledge about learning that enables us to understand and improve the performance of learning algorithms.     

Learning concept descriptions with typed evolutionary programming

Examples and concepts in traditional concept learning tasks are represented with the attribute-value language. While enabling efficient implementations, we argue that such propositional representation is inadequate when data is rich in structure. This paper describes STEPS, a strongly-typed evolutionary programming system designed to induce concepts from structured data. STEPS higher-order logic representation language enhances expressiveness, while the use of evolutionary computation dampens the effects of the corresponding explosion of the search space. Results on the PTE2 challenge, a major real-world knowledge discovery application from the molecular biology domain, demonstrate promise.     

Implicit affinity networks and social capital

Social networks are typically constructed based on explicit and well-defined relationships among individuals. In this paper, we describe another class of social networks, known as implicit affinity networks, where links are implicit in the patterns of natural affinities among individuals. An effective mathematical formulation of social capital based on implicit and explicit connections is given. Results with two Web communities, one focused on people’s interests and one focused on people’s blogs, exhibit rich dynamics and show interesting patterns of community evolution.     

Behavior-based clustering and analysis of interestingness measures for association rule mining

A number of studies, theoretical, empirical, or both, have been conducted to provide insight into the properties and behavior of interestingness measures for association rule mining. While each has value in its own right, most are either limited in scope or, more importantly, ignore the purpose for which interestingness measures are intended, namely the ultimate ranking of discovered association rules. This paper, therefore, focuses on an analysis of the rule-ranking behavior of 61 well-known interestingness measures tested on the rules generated from 110 different datasets. By clustering based on ranking behavior, we highlight, and formally prove, previously unreported equivalences among interestingness measures. We also show that there appear to be distinct clusters of interestingness measures, but that there remain differences among clusters, confirming that domain knowledge is essential to the selection of an appropriate interestingness measure for a particular task and business objective.     

Metalearning and Algorithm Selection: progress,state of the art and introduction to the 2018 Special Issue

This article serves as an introduction to the Special Issue on Metalearning and Algorithm Selection. The introduction is divided into two parts. In the the first section, we give an overview of how the field of metalearning has evolved in the last 1–2 decades and mention how some of the papers in this special issue fit in. In the second section, we discuss the contents of this special issue. We divide the papers into thematic subgroups, provide information about each subgroup, as well as about the individual papers. Our main aim is to highlight how the papers selected for this special issue contribute to the field of metalearning.     

An instance level analysis of data complexity

Most data complexity studies have focused on characterizing the complexity of the entire data set and do not provide information about individual instances. Knowing which instances are misclassified and understanding why they are misclassified and how they contribute to data set complexity can improve the learning process and could guide the future development of learning algorithms and data analysis methods. The goal of this paper is to better understand the data used in machine learning problems by identifying and analyzing the instances that are frequently misclassified by learning algorithms that have shown utility to date and are commonly used in practice. We identify instances that are hard to classify correctly (instance hardness) by classifying over 190,000 instances from 64 data sets with 9 learning algorithms. We then use a set of hardness measures to understand why some instances are harder to classify correctly than others. We find that class overlap is a principal contributor to instance hardness. We seek to integrate this information into the training process to alleviate the effects of class overlap and present ways that instance hardness can be used to improve learning.     

A method for computing lexical semantic distance using linear functionals

This paper presents a novel, knowledge-based method for measuring semantic similarity in support of applications aimed at organizing and retrieving relevant textual information. We show how a quantitative context may be established for what is essentially qualitative in nature by effecting a topological transformation of the lexicon into a metric space where distance is well-defined. We illustrate the technique with a simple example and report on promising experimental results with a significant word similarity problem.