Introduction to the Special Issue on Meta-Learning

Abstract

Abstract. 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.

Meta-learning to optimize the number of hidden nodes of MLP networks trained by Extreme Learning Machine algorithm

“…Meta-learning can be defined as the process of automatic acquisition of knowledge which relates the performance of learning algorithms with the characteristics of the learning problems [7]. Figure 1 presents the meta-learning approach.…”

“…Meta-learning is an area that tries to extrapolate the knowledge acquired in past problems to solve new problems. Meta-learning can be defined as the process of automatic acquisition of knowledge that relates the performance of learning algorithms to the characteristics of problems [7]. The knowledge in meta-learning is acquired from a set of metaexamples.…”

AMORI: A Metric-based One Rule Inducer

“…Researchers in the field of meta-learning have pointed out the importance of going beyond the engineering goal of producing more accurate learners to the scientific goal of understanding learning behavior [11]. We believe that it is possible to move closer to this goal by conducting research into some of the core concepts of metric-based learning.…”

Hidden Markov Models and Text Classifiers for Information Extraction on Semi-Structured Texts

“…Our approach is strongly related to the Stacked Generalization technique [29], which consists of training a new classifier using as input the output provided by other classifiers, in a kind of meta-learning [12]. However, our strategy is not to combine the output of different classifiers, but rather to use an HMM to refine the classification delivered by a single classifier for all input fragments.…”