Showing papers on "Algorithmic learning theory published in 2006"

Can machine learning be secure

Abstract: Machine learning systems offer unparalled flexibility in dealing with evolving input in a variety of applications, such as intrusion detection systems and spam e-mail filtering. However, machine learning algorithms themselves can be a target of attack by a malicious adversary. This paper provides a framework for answering the question, "Can machine learning be secure?" Novel contributions of this paper include a taxonomy of different types of attacks on machine learning techniques and systems, a variety of defenses against those attacks, a discussion of ideas that are important to security for machine learning, an analytical model giving a lower bound on attacker's work function, and a list of open problems.

The Mathematics of Learning: Dealing With Data

Abstract: Learning is key to developing systems tailored to a broad range of data analysis and information extraction tasks. We outline the mathematical foundations of learning theory and describe a key algorithm of it.

Multi-Instance Multi-Label Learning with Application to Scene Classification

Abstract: In this paper, we formalize multi-instance multi-label learning, where each training example is associated with not only multiple instances but also multiple class labels Such a problem can occur in many real-world tasks, eg an image usually contains multiple patches each of which can be described by a feature vector, and the image can belong to multiple categories since its semantics can be recognized in different ways We analyze the relationship between multi-instance multi-label learning and the learning frameworks of traditional supervised learning, multi-instance learning and multi-label learning Then, we propose the MIMLBOOST and MIMLSVM algorithms which achieve good performance in an application to scene classification

Machine Learning for Detection and Diagnosis of Disease

Abstract: Machine learning offers a principled approach for developing sophisticated, automatic, and objective algorithms for analysis of high-dimensional and multimodal biomedical data. This review focuses on several advances in the state of the art that have shown promise in improving detection, diagnosis, and therapeutic monitoring of disease. Key in the advancement has been the development of a more in-depth understanding and theoretical analysis of critical issues related to algorithmic construction and learning theory. These include trade-offs for maximizing generalization performance, use of physically realistic constraints, and incorporation of prior knowledge and uncertainty. The review describes recent developments in machine learning, focusing on supervised and unsupervised linear methods and Bayesian inference, which have made significant impacts in the detection and diagnosis of disease in biomedicine. We describe the different methodologies and, for each, provide examples of their application to specific domains in biomedical diagnostics.

Statistical Learning Within and Between Modalities Pitting Abstract Against Stimulus-Specific Representations

Abstract: When learners encode sequential patterns and generalize their knowledge to novel instances, are they relying on abstract or stimulus-specific representations? Research on artificial grammar learning (AGL) has shown transfer of learning from one stimulus set to another, and such findings have encouraged the view that statistical learning is mediated by abstract representations that are independent of the sense modality or perceptual features of the stimuli. Using a novel modification of the standard AGL paradigm, we obtained data to the contrary. These experiments pitted abstract processing against stimulus-specific learning. The findings show that statistical learning results in knowledge that is stimulus-specific rather than abstract. They show furthermore that learning can proceed in parallel for multiple input streams along separate perceptual dimensions or sense modalities. We conclude that learning sequential structure and generalizing to novel stimuli inherently involve learning mechanisms that are ...

Human Neural Learning Depends on Reward Prediction Errors in the Blocking Paradigm

Abstract: Learning occurs when an outcome deviates from expectation (prediction error). According to formal learning theory, the defining paradigm demonstrating the role of prediction errors in learning is the blocking test. Here, a novel stimulus is blocked from learning when it is associated with a fully predicted outcome, presumably because the occurrence of the outcome fails to produce a prediction error. We investigated the role of prediction errors in human reward-directed learning using a blocking paradigm and measured brain activation with functional magnetic resonance imaging. Participants showed blocking of behavioral learning with juice rewards as predicted by learning theory. The medial orbitofrontal cortex and the ventral putamen showed significantly lower responses to blocked, compared with nonblocked, reward-predicting stimuli. In reward-predicting control situations, deactivations in orbitofrontal cortex and ventral putamen occurred at the time of unpredicted reward omissions. Responses in discrete parts of orbitofrontal cortex correlated with the degree of behavioral learning during, and after, the learning phase. These data suggest that learning in primary reward structures in the human brain correlates with prediction errors in a manner that complies with principles of formal learning theory.

Analogical Processes in Language Learning

Abstract: The acquisition of language has long stood as a challenge to general learning accounts, leading many theorists to propose domain-specific knowledge and processes to explain language acquisition. Here we review evidence that analogical comparison is instrumental in language learning, suggesting a larger role for general learning processes in the acquisition of language.

The Interplay of Optimization and Machine Learning Research

Abstract: The fields of machine learning and mathematical programming are increasingly intertwined. Optimization problems lie at the heart of most machine learning approaches. The Special Topic on Machine Learning and Large Scale Optimization examines this interplay. Machine learning researchers have embraced the advances in mathematical programming allowing new types of models to be pursued. The special topic includes models using quadratic, linear, second-order cone, semi-definite, and semi-infinite programs. We observe that the qualities of good optimization algorithms from the machine learning and optimization perspectives can be quite different. Mathematical programming puts a premium on accuracy, speed, and robustness. Since generalization is the bottom line in machine learning and training is normally done off-line, accuracy and small speed improvements are of little concern in machine learning. Machine learning prefers simpler algorithms that work in reasonable computational time for specific classes of problems. Reducing machine learning problems to well-explored mathematical programming classes with robust general purpose optimization codes allows machine learning researchers to rapidly develop new techniques. In turn, machine learning presents new challenges to mathematical programming. The special issue include papers from two primary themes: novel machine learning models and novel optimization approaches for existing models. Many papers blend both themes, making small changes in the underlying core mathematical program that enable the develop of effective new algorithms.

An Ontology-Based Framework for Bridging Learning Design and Learning Content

Abstract: The paper describes an ontology-based framework for bridging learning design and learning object content. In present solutions, researchers have proposed conceptual models and developed tools for both of those subjects, but without detailed discussions of how they can be used together. In this paper we advocate the use of ontologies to explicitly specify all learning designs, learning objects, and the relations between them, and show how this use of ontologies can result in more effective (semi-)automatic tools and services that increase the level of reusability. We first define a three-part conceptual model that introduces an intermediate level between learning design and learning objects called the learning object context. We then use ontologies to facilitate the representation of these concepts: LOCO is a new ontology based on IMS-LD, ALOCoM is an existing ontology for learning objects, and LOCO-Cite is a new ontology for the learning object contextual model. We conclude by showing the applicability of the proposed framework in a use case study.

Learning Design Based on Graphical Knowledge-Modelling.

Abstract: This chapter states and explains that a Learning Design is the result of a knowledge engineering process where knowledge and competencies, learning design and delivery models are constructed in an integrated framework. We present a general graphical language and a knowledge editor that has been adapted to support the construction of learning designs compliant with the IMS-LD specification. We situate LD within our taxonomy of knowledge models as a multi-actor collaborative system. We move up one step in the abstraction scale, showing that the process of constructing learning designs can itself be viewed as a unit-of-learning (or a “unit-of-design”): designers can be seen as learning by constructing learning designs, individually, in teams and with staff support. This viewpoint enables us to discuss and compare various “design plays”. Further, the issue of representing knowledge, cognitive skills and competencies is addressed. The association between these “content” models and learning design components can guide the construction of learning designs and help to classify them in repositories of LD templates.

Active learning in the non-realizable case

Abstract: Most of the existing active learning algorithms are based on the realizability assumption: The learner's hypothesis class is assumed to contain a target function that perfectly classifies all training and test examples. This assumption can hardly ever be justified in practice. In this paper, we study how relaxing the realizability assumption affects the sample complexity of active learning. First, we extend existing results on query learning to show that any active learning algorithm for the realizable case can be transformed to tolerate random bounded rate class noise. Thus, bounded rate class noise adds little extra complications to active learning, and in particular exponential label complexity savings over passive learning are still possible. However, it is questionable whether this noise model is any more realistic in practice than assuming no noise at all. Our second result shows that if we move to the truly non-realizable model of statistical learning theory, then the label complexity of active learning has the same dependence Ω(1/e2) on the accuracy parameter e as the passive learning label complexity. More specifically, we show that under the assumption that the best classifier in the learner's hypothesis class has generalization error at most β>0, the label complexity of active learning is Ω(β2/e2log(1/δ)), where the accuracy parameter e measures how close to optimal within the hypothesis class the active learner has to get and δ is the confidence parameter. The implication of this lower bound is that exponential savings should not be expected in realistic models of active learning, and thus the label complexity goals in active learning should be refined.

An Empirical Study of the Behavior of Active Learning for Word Sense Disambiguation

Abstract: This paper shows that two uncertainty-based active learning methods, combined with a maximum entropy model, work well on learning English verb senses. Data analysis on the learning process, based on both instance and feature levels, suggests that a careful treatment of feature extraction is important for the active learning to be useful for WSD. The overfitting phenomena that occurred during the active learning process are identified as classic overfitting in machine learning based on the data analysis.

Socially guided machine learning

Abstract: Social interaction will be key to enabling robots and machines in general to learn new tasks from ordinary people (not experts in robotics or machine learning). Everyday people who need to teach their machines new things will find it natural for to rely on their interpersonal interaction skills. This thesis provides several contributions towards the understanding of this Socially Guided Machine Learning scenario. While the topic of human input to machine learning algorithms has been explored to some extent, prior works have not gone far enough to understand what people will try to communicate when teaching a machine and how algorithms and learning systems can be modified to better accommodate a human partner. Interface techniques have been based on intuition and assumptions rather than grounded in human behavior, and often techniques are not demonstrated or evaluated with everyday people. Using a computer game, Sophie's Kitchen, an experiment with human subjects provides several insights about how people approach the task of teaching a machine. In particular, people want to direct and guide an agent's exploration process, they quickly use the behavior of the agent to infer a mental model of the learning process, and they utilize positive and negative feedback in asymmetric ways. Using a robotic platform, Leonardo, and 200 people in follow-up studies of modified versions of the Sophie's Kitchen game, four research themes are developed. The use of human guidance in a machine learning exploration can be successfully incorporated to improve learning performance. Novel learning approaches demonstrate aspects of goal-oriented learning. The transparency of the machine learner can have significant effects on the nature of the instruction received from the human teacher, which in turn positively impacts the learning process. Utilizing asymmetric interpretations of positive and negative feedback from a human partner, can result in a more efficient and robust learning experience. (Copies available exclusively from MIT Libraries, Rm. 14-0551, Cambridge, MA 02139-4307. Ph. 617-253-5668; Fax 617-253-1690.)

E-learning: a model and process proposal

Abstract: The advent of the new technologies has brought the proliferation of artificial environments in distance and blended learning Most of the efforts have been addressed to the mere 'transfer' of paper and pen Thus, we proposed the following theoretical reflection on the learning processes: studying classical didactics that have been revised from the point-of-view of distance learning, individuating the new actors interacting in the didactics, and deducing a possible learning model The theoretical reflection, presented in its model and conceptual aspects, finds its application in the realisation of an e-learning platform, IWT The advantages of the state-of-the-art learning are also shown in this paper However, this work is essentially limited to a general positioning of e-learning and to a presentation motivated by the choices we have made in the theoretical, architectural, functional, and technological fields

Chapter 18 Agent Learning Representation: Advice on Modelling Economic Learning

Abstract: This chapter presents an overview of the existing learning models in the economic literature. Furthermore, it discusses the choice of models that should be used under various circumstances and how adequate learning models can be chosen in simulation approaches. It gives advice for using the many existing models and selecting the appropriate model for each application.

IMS Learning Design Version 1.0 Final Specification

Abstract: The IMS Learning Design specification supports the use of a wide range of pedagogies in online learning Rather than attempting to capture the specifics of many pedagogies, it does this by providing a generic and flexible language This language is designed to enable many different pedagogies to be expressed The approach has the advantage over alternatives in that only one set of learning design and runtime tools then need to be implemented in order to support the desired wide range of pedagogies The language was originally developed at the Open University of the Netherlands (OUNL), after extensive examination and comparison of a wide range of pedagogical approaches and their associated learning activities, and several iterations of the developing language to obtain a good balance between generality and pedagogic expressiveness

Semi-Supervised Learning - A Statistical Physics Approach

Abstract: We present a novel approach to semisupervised learning which is based on statistical physics. Most of the former work in the field of semi-supervised learning classifies the points by minimizing a certain energy function, which corresponds to a minimal k-way cut solution. In contrast to these methods, we estimate the distribution of classifications, instead of the sole minimal k-way cut, which yields more accurate and robust results. Our approach may be applied to all energy functions used for semi-supervised learning. The method is based on sampling using a Multicanonical Markov chain Monte-Carlo algorithm, and has a straightforward probabilistic interpretation, which allows for soft assignments of points to classes, and also to cope with yet unseen class types. The suggested approach is demonstrated on a toy data set and on two real-life data sets of gene expression.

Mathematics based on incremental learning: excluded middle and inductive inference

Abstract: Learning theoretic aspects of mathematics and logic have been studied by many authors. They study how mathematical and logical objects are algorithmically "learned" (inferred) from finite data. Although they study mathematical objects, the objective of the studies is learning. In this paper, a mathematics whose foundation itself is learning theoretic will be introduced. It is called Limit-Computable Mathematics. It was originally introduced as a means for "Proof Animation", which is expected to make interactive formal proof development easier. Although the original objective was not learning theoretic at all, learning theory is indispensable for our research. It suggests that logic and learning theory are related in a still unknown but deep new way.

Learning Transliteration Lexicons from the Web

Abstract: This paper presents an adaptive learning framework for Phonetic Similarity Modeling (PSM) that supports the automatic construction of transliteration lexicons. The learning algorithm starts with minimum prior knowledge about machine transliteration, and acquires knowledge iteratively from the Web. We study the active learning and the unsupervised learning strategies that minimize human supervision in terms of data labeling. The learning process refines the PSM and constructs a transliteration lexicon at the same time. We evaluate the proposed PSM and its learning algorithm through a series of systematic experiments, which show that the proposed framework is reliably effective on two independent databases.

Chapter 8 Machine Learning in Computational Chemistry

Abstract: Publisher Summary This chapter provides an overview of machine learning techniques that have recently appeared in the computational chemistry literature. The natural fit between machine learning and pharmaceutical research leads to the common utilization of learning algorithms to construct quantitative structure activity relationships (QSAR). Until recently, the application of machine learning methods is primarily the domain of specialists in the artificial intelligence or statistics fields. Early machine learning software requires an intimate knowledge of the algorithms and a familiarity with specialized programming languages, such as Prolog or Lisp. The availability of easily accessible, high-quality software has led to the widespread adoption of machine learning among computational chemists. Although modern machine learning techniques require minimal parameter adjustment to achieve reasonable results, it is inappropriate to treat these techniques as black boxes. As with other QSAR techniques, model quality and applicability must be judiciously assessed.

Learning Pronunciation Dictionaries: Language Complexity and Word Selection Strategies

Abstract: The speed with which pronunciation dictionaries can be bootstrapped depends on the efficiency of learning algorithms and on the ordering of words presented to the user. This paper presents an active-learning word selection strategy that is mindful of human limitations. Learning rates approach that of an oracle system that knows the final LTS rule set.

Learning automata as a basis for multi agent reinforcement learning

Abstract: In this paper we summarize some important theoretical results from the domain of Learning Automata. We start with single stage, single agent learning schema's, and gradually extend the setting to multistage multi agent systems. We argue that the theory of Learning Automata is an ideal basis to build multi agent learning algorithms.

Multi-Instance Learning from Supervised View

Abstract: In multi-instance learning, the training set comprises labeled bags that are composed of unlabeled instances, and the task is to predict the labels of unseen bags. This paper studies multi-instance learning from the view of supervised learning. First, by analyzing some representative learning algorithms, this paper shows that multi-instance learners can be derived from supervised learners by shifting their focuses from the discrimination on the instances to the discrimination on the bags. Second, considering that ensemble learning paradigms can effectively enhance supervised learners, this paper proposes to build multi-instance ensembles to solve multi-instance problems. Experiments on a real-world benchmark test show that ensemble learning paradigms can significantly enhance multi-instance learners.

Perceptual learning and abstraction in machine learning: an application to autonomous robotics

Abstract: This paper deals with the possible benefits of perceptual learning in artificial intelligence. On the one hand, perceptual learning is more and more studied in neurobiology and is now considered as an essential part of any living system. In fact, perceptual learning and cognitive learning are both necessary for learning and often depend on each other. On the other hand, many works in machine learning are concerned with "abstraction" in order to reduce the amount of complexity related to some learning tasks. In the abstraction framework, perceptual learning can be seen as a specific process that learns how to transform the data before the traditional learning task itself takes place. In this paper, we argue that biologically inspired perceptual learning mechanisms could be used to build efficient low-level abstraction operators that deal with real-world data. To illustrate this, we present an application where perceptual-learning-inspired metaoperators are used to perform an abstraction on an autonomous robot visual perception. The goal of this work is to enable the robot to learn how to identify objects it encounters in its environment.

Predictions as statements and decisions

Abstract: Prediction is a complex notion, and different predictors (such as people, computer programs, and probabilistic theories) can pursue very different goals. In this talk I will review some popular kinds of prediction and argue that the theory of competitive on-line learning can benefit from the kinds of prediction that are now foreign to it.

On supporting the process of learning design through planners

Abstract: Recent standardization in learning technology has resulted in a model of activity-based learning designs called IMS LD that provides a common framework for expressing any kind of activity-based learning program. This emphasis on designing learning programs based on assembling activities and resources in turn allows the application of Artificial Intelligence (AI) techniques to help in the process of design. However, learning design is a problem of open rationality, and the applicability of computational techniques is a controversial matter in itself. This paper explores the applicability of common planners to the (partial) automation of learning design, and provides the general guidelines for the design of pedagogical designer agents. Even though such designers can not provide unique or deterministic solutions - due to inherent characteristics of the problem - they can be equipped with different "rationalities" about human learning, eventually leading to new insights in the conceptions of learning, gained through the observation of computational models that embody the main principles and action guidelines of pedagogical design approaches.

Modes of learning in the use of a computer-based business simulation game

Abstract: This paper discusses the role of computer based simulations in business education. It examines the learning approaches adopted by students using a simulation game. Results inform the development of causal-loop diagrams capturing representations of zero, single and double-loop learning within the study context. Actions are proposed to maximise the effectiveness of this form of learning technology.

Alleviating Catastrophic Forgetting via Multi-Objective Learning

Abstract: Handling catastrophic forgetting is an interesting and challenging topic in modeling the memory mechanisms of the human brain using machine learning models. From a more general point of view, catastrophic forgetting reflects the stability-plasticity dilemma, which is one of the several dilemmas to be addressed in learning systems: to retain the stored memory while learning new information. Different to the existing approaches, we introduce a Pareto-optimality based multi-objective learning framework for alleviating catastrophic learning. Compared to the single-objective learning methods, multi-objective evolutionary learning with the help of pseudo-rehearsal is shown to be more promising in dealing with the stability-plasticity dilemma.