Learning in the Absence of Negative Examples

Machine Learning, 1991

This paper is aimed at showing the benefits obtained by explicitly introducing a priori control knowledge into the inductive process. The starting point is Michalski's Induce system, which has been modified and augmented. Although the basic philosophy has been changed as little as possible, Induce has been radically modified from the algorithmic point of view, resulting in the new learning system Rigel. The main ideas taken from Induce are the sequential learning of descriptions of each concept against all the others, the Covering algorithm, the Star definition, and the VL 2 representation language. The modifications consist of a new way of computing the Star, the use of a separate body of heuristic knowledge to strongly direct the search, the implementation of a larger subset of the VL 2 language, a reasoned way of selecting the seed, and the use of rules to evaluate the worthiness of the inductive assertions. The effectiveness of Rigel has been tested both on artificial and on real-world case studies.

Positive and unlabelled learning (PU learning) has been investigated to deal with the situation where only the positive examples and the unlabelled examples are available. Most of the previous works focus on identifying some negative examples from the unlabelled data, so that the supervised learning methods can be applied to build a classifier. However, for the remaining unlabelled data, which can not be explicitly identified as positive or negative (we call them ambiguous examples), they either exclude them from the training phase or simply enforce them to either class. Consequently, their performance may be constrained. This paper proposes a novel approach, called similarity-based PU learning (SPUL) method, by associating the ambiguous examples with two similarity weights, which indicate the similarity of an ambiguous example towards the positive class and the negative class, respectively. The local similarity-based and global similarity-based mechanisms are proposed to generate t...

1992

We propose an instance-based learning algorithm named IBRL3 which acquires some kind of control rules not to fail in a dynamic environment, and we examine its performance via application to the cart-pole balancing problem. In this algorithm, a tuple of input, output and preference value of each execution cycle are stored in memory verbatim, and the action of each cycle is decided by retrieving the nearest neighbor of the current input data. The number of stored instances is reduced by replacing the nearest but less reliable instance by new one. Experimental results of computer simulation show that IBRL3 is robust for distinct settings of parameters and for noisy environments, and is efficient enough to apply to some kinds of real-time control problems.

2010

An investigation of rule learning processes that allow the inclusion of negated features is described. The objective is to establish whether the use of negation in inductive rule learning systems is effective with respect to classification. This paper seeks to answer this question by considering two issues relevant to such systems; feature identification and rule refinement.

This paper introduces a new type of intelligent agent called a constructive induction-based learning agent (CILA). This agent differs from other adaptive agents because it has the ability to not only learn how to assist a user in some task, but also to incrementally adapt its knowledge representation space to better fit the given learning task. The agent's ability to autonomously make problem-oriented modifications to the originally given representation space is due to its constructive induction (CI) learning method. Selective induction (SI) learning methods, and agents based on these methods, rely on a good representation space. A good representation space has no misclassification noise, inter-correlated attributes or irrelevant attributes. Our proposed CILA has methods for overcoming all of these problems. In agent domains with poor representations, the CIbased learning agent will learn more accurate rules and be more useful than an SI-based learning agent. This paper gives an...

2010

Bootstrap Learning (BL) is a new machine learning paradigm that seeks to build an electronic student that can learn using natural instruction provided by a human teacher and by bootstrapping on previously learned concepts. In our setting, the teacher provides (very few) examples and some advice about the task at hand using a natural instruction interface. To address this task, we use our Inductive Logic Programming system called WILL to translate the natural instruction into first-order logic. We present approaches to the various challenges BL raises, namely automatic translation of domain knowledge and instruction into an ILP problem and the automation of ILP runs across different tasks and domains, which we address using a multi-layered approach. We demonstrate that our system is able to learn effectively in over fifty different lessons across three different domains without any humanperformed parameter tuning between tasks.

2010

We developed a new model for iList, our system that helps students learn linked list. The model is automatically extracted from past student data, and allows iList to track students' problem-solving behavior in order to provide targeted feedback. We evaluated the new model both intrinsically and extrinsically. We show that the model can match most student actions after a relatively small sequence of observations, and that iList can effectively use the new student tracker to provide feedback and help students learn.

2006

Several upgrades of Attribute-Value learning to Inductive Logic Programming have been proposed and used successfully. However, the Top-Down Data-Driven strategy, popularised by the AQ family, has not yet been transferred to ILP: if the idea of reducing the hypothesis space by covering a seed example is utilised with systems like PROGOL, Aleph or MIO, these systems do not benefit from the associated data-driven specialisation operator. This operator is given an incorrect hypothesis h and a covered negative example e and outputs a set of hypotheses more specific than h and correct wrt e. This refinement operator is very valuable considering heuristic search problems ILP systems may encounter when crossing plateaus in relational search spaces. In this paper, we present the data-driven strategy of AQ, in terms of a lgg-based change of representation of negative examples given a positive seed example, and show how it can be extended to ILP. We evaluate a basic implementation of AQ in the system Propal on a number of benchmark ILP datasets.

1997

This paper describes the design, implementation and evaluation of an agent that learns to automate repetitive tasks within the human-computer interface. In contrast to most Artificial Intelligence projects, the design centers on a user study, with a human-simulated agent used to discover the interactions that people find natural. This study shows the ways in which users instinctively communicate via "hints," or partially-specified, ambiguous, instructions. Hints may be communicated using speech, by pointing, or by selecting from menus. We develop a taxonomy of such instructions. The implementation demonstrates that computers can learn from examples and ambiguous hints.

1983

Research in the area of learning structural descriptions from examples is reviewed, giving primary attention to methods of learning characteristic descrip tions of single concepts. In particular, we examine methods for finding the maximally-specific conjunctive generalizations (MSC-generalizations) that cover all of the training examples of a given concept. Various important aspects of structural learning in general are examined, and several criteria for evaluating structural learning methods are presented. Briefly, these criteria include (i) ade quacy of the representation language, (ii) generalization rules employed, computational efficiency, and (iv) flexibility and extensibility. Selected learning methods developed by Buchanan, et al., Hayes-Roth, Vere, Winston, and the authors are analyzed according to these criteria. Finally, some goals are sug gested for future research.