Abstract: The goal of machine learning is to program computers to use example data or past experience to solve a given problem. Many successful applications of machine learning exist already, including systems that analyze past sales data to predict customer behavior, optimize robot behavior so that a task can be completed using minimum resources, and extract knowledge from bioinformatics data. Introduction to Machine Learning is a comprehensive textbook on the subject, covering a broad array of topics not usually included in introductory machine learning texts. In order to present a unified treatment of machine learning problems and solutions, it discusses many methods from different fields, including statistics, pattern recognition, neural networks, artificial intelligence, signal processing, control, and data mining. All learning algorithms are explained so that the student can easily move from the equations in the book to a computer program. The text covers such topics as supervised learning, Bayesian decision theory, parametric methods, multivariate methods, multilayer perceptrons, local models, hidden Markov models, assessing and comparing classification algorithms, and reinforcement learning. New to the second edition are chapters on kernel machines, graphical models, and Bayesian estimation; expanded coverage of statistical tests in a chapter on design and analysis of machine learning experiments; case studies available on the Web (with downloadable results for instructors); and many additional exercises. All chapters have been revised and updated. Introduction to Machine Learning can be used by advanced undergraduates and graduate students who have completed courses in computer programming, probability, calculus, and linear algebra. It will also be of interest to engineers in the field who are concerned with the application of machine learning methods. Adaptive Computation and Machine Learning series

Abstract: We consider the semi-supervised learning problem, where a decision rule is to be learned from labeled and unlabeled data. In this framework, we motivate minimum entropy regularization, which enables to incorporate unlabeled data in the standard supervised learning. Our approach includes other approaches to the semi-supervised problem as particular or limiting cases. A series of experiments illustrates that the proposed solution benefits from unlabeled data. The method challenges mixture models when the data are sampled from the distribution class spanned by the generative model. The performances are definitely in favor of minimum entropy regularization when generative models are misspecified, and the weighting of unlabeled data provides robustness to the violation of the "cluster assumption". Finally, we also illustrate that the method can also be far superior to manifold learning in high dimension spaces.

Abstract: Past empirical work has shown that learning multiple related tasks from data simultaneously can be advantageous in terms of predictive performance relative to learning these tasks independently. In this paper we present an approach to multi--task learning based on the minimization of regularization functionals similar to existing ones, such as the one for Support Vector Machines (SVMs), that have been successfully used in the past for single--task learning. Our approach allows to model the relation between tasks in terms of a novel kernel function that uses a task--coupling parameter. We implement an instance of the proposed approach similar to SVMs and test it empirically using simulated as well as real data. The experimental results show that the proposed method performs better than existing multi--task learning methods and largely outperforms single--task learning using SVMs.

Abstract: Principal component analysis (PCA) is a widely used statistical technique for unsupervised dimension reduction. K-means clustering is a commonly used data clustering for performing unsupervised learning tasks. Here we prove that principal components are the continuous solutions to the discrete cluster membership indicators for K-means clustering. New lower bounds for K-means objective function are derived, which is the total variance minus the eigenvalues of the data covariance matrix. These results indicate that unsupervised dimension reduction is closely related to unsupervised learning. Several implications are discussed. On dimension reduction, the result provides new insights to the observed effectiveness of PCA-based data reductions, beyond the conventional noise-reduction explanation that PCA, via singular value decomposition, provides the best low-dimensional linear approximation of the data. On learning, the result suggests effective techniques for K-means data clustering. DNA gene expression and Internet newsgroups are analyzed to illustrate our results. Experiments indicate that the new bounds are within 0.5-1.5% of the optimal values.

Abstract: Learning general functional dependencies is one of the main goals in machine learning. Recent progress in kernel-based methods has focused on designing flexible and powerful input representations. This paper addresses the complementary issue of problems involving complex outputs such as multiple dependent output variables and structured output spaces. We propose to generalize multiclass Support Vector Machine learning in a formulation that involves features extracted jointly from inputs and outputs. The resulting optimization problem is solved efficiently by a cutting plane algorithm that exploits the sparseness and structural decomposition of the problem. We demonstrate the versatility and effectiveness of our method on problems ranging from supervised grammar learning and named-entity recognition, to taxonomic text classification and sequence alignment.

Abstract: An approach to the online learning of Takagi-Sugeno (TS) type models is proposed in the paper. It is based on a novel learning algorithm that recursively updates TS model structure and parameters by combining supervised and unsupervised learning. The rule-base and parameters of the TS model continually evolve by adding new rules with more summarization power and by modifying existing rules and parameters. In this way, the rule-base structure is inherited and up-dated when new data become available. By applying this learning concept to the TS model we arrive at a new type adaptive model called the Evolving Takagi-Sugeno model (ETS). The adaptive nature of these evolving TS models in combination with the highly transparent and compact form of fuzzy rules makes them a promising candidate for online modeling and control of complex processes, competitive to neural networks. The approach has been tested on data from an air-conditioning installation serving a real building. The results illustrate the viability and efficiency of the approach. The proposed concept, however, has significantly wider implications in a number of fields, including adaptive nonlinear control, fault detection and diagnostics, performance analysis, forecasting, knowledge extraction, robotics, behavior modeling.

Abstract: In this paper, we identify two issues involved in developing an automated feature subset selection algorithm for unlabeled data: the need for finding the number of clusters in conjunction with feature selection, and the need for normalizing the bias of feature selection criteria with respect to dimension. We explore the feature selection problem and these issues through FSSEM (Feature Subset Selection using Expectation-Maximization (EM) clustering) and through two different performance criteria for evaluating candidate feature subsets: scatter separability and maximum likelihood. We present proofs on the dimensionality biases of these feature criteria, and present a cross-projection normalization scheme that can be applied to any criterion to ameliorate these biases. Our experiments show the need for feature selection, the need for addressing these two issues, and the effectiveness of our proposed solutions.

Abstract: SUSTAIN (Supervised and Unsupervised STratified Adaptive Incremental Network) is a model of how humans learn categories from examples. SUSTAIN initially assumes a simple category structure. If simple solutions prove inadequate and SUSTAIN is confronted with a surprising event (e.g., it is told that a bat is a mammal instead of a bird), SUSTAIN recruits an additional cluster to represent the surprising event. Newly recruited clusters are available to explain future events and can themselves evolve into prototypes-attractors-rules. SUSTAIN's discovery of category substructure is affected not only by the structure of the world but by the nature of the learning task and the learner's goals. SUSTAIN successfully extends category learning models to studies of inference learning, unsupervised learning, category construction, and contexts in which identification learning is faster than classification learning.

Abstract: The paper is concerned with two-class active learning. While the common approach for collecting data in active learning is to select samples close to the classification boundary, better performance can be achieved by taking into account the prior data distribution. The main contribution of the paper is a formal framework that incorporates clustering into active learning. The algorithm first constructs a classifier on the set of the cluster representatives, and then propagates the classification decision to the other samples via a local noise model. The proposed model allows to select the most representative samples as well as to avoid repeatedly labeling samples in the same cluster. During the active learning process, the clustering is adjusted using the coarse-to-fine strategy in order to balance between the advantage of large clusters and the accuracy of the data representation. The results of experiments in image databases show a better performance of our algorithm compared to the current methods.

Abstract: Clustering is a common unsupervised learning technique used to discover group structure in a set of data. While there exist many algorithms for clustering, the important issue of feature selection, that is, what attributes of the data should be used by the clustering algorithms, is rarely touched upon. Feature selection for clustering is difficult because, unlike in supervised learning, there are no class labels for the data and, thus, no obvious criteria to guide the search. Another important problem in clustering is the determination of the number of clusters, which clearly impacts and is influenced by the feature selection issue. In this paper, we propose the concept of feature saliency and introduce an expectation-maximization (EM) algorithm to estimate it, in the context of mixture-based clustering. Due to the introduction of a minimum message length model selection criterion, the saliency of irrelevant features is driven toward zero, which corresponds to performing feature selection. The criterion and algorithm are then extended to simultaneously estimate the feature saliencies and the number of clusters.

Abstract: Can we detect low dimensional structure in high dimensional data sets of images and video? The problem of dimensionality reduction arises often in computer vision and pattern recognition. In this paper, we propose a new solution to this problem based on semidefinite programming. Our algorithm can be used to analyze high dimensional data that lies on or near a low dimensional manifold. It overcomes certain limitations of previous work in manifold learning, such as Isomap and locally linear embedding. We illustrate the algorithm on easily visualized examples of curves and surfaces, as well as on actual images of faces, handwritten digits, and solid objects.

Abstract: A key problem in learning multiple objects from unlabeled images is that it is a priori impossible to tell which part of the image corresponds to each individual object, and which part is irrelevant clutter which is not associated to the objects. We investigate empirically to what extent pure bottom-up attention can extract useful information about the location, size and shape of objects from images and demonstrate how this information can be utilized to enable unsupervised learning of objects from unlabeled images. Our experiments demonstrate that the proposed approach to using bottom-up attention is indeed useful for a variety of applications.

Abstract: Outlier detection is a fundamental issue in data mining, specifically in fraud detection, network intrusion detection, network monitoring, etc. SmartSifter is an outlier detection engine addressing this problem from the viewpoint of statistical learning theory. This paper provides a theoretical basis for SmartSifter and empirically demonstrates its effectiveness. SmartSifter detects outliers in an on-line process through the on-line unsupervised learning of a probabilistic model (using a finite mixture model) of the information source. Each time a datum is input SmartSifter employs an on-line discounting learning algorithm to learn the probabilistic model. A score is given to the datum based on the learned model with a high score indicating a high possibility of being a statistical outlier. The novel features of SmartSifter are: (1) it is adaptive to non-stationary sources of datas (2) a score has a clear statistical/information-theoretic meanings (3) it is computationally inexpensives and (4) it can handle both categorical and continuous variables. An experimental application to network intrusion detection shows that SmartSifter was able to identify data with high scores that corresponded to attacks, with low computational costs. Further experimental application has identified a number of meaningful rare cases in actual health insurance pathology data from Australia's Health Insurance Commission.

Abstract: This article addresses the representation of numerical information conveyed by nonsymbolic and symbolic stimuli. In a first simulation study, we show how number-selective neurons develop when an initially uncommitted neural network is given nonsymbolic stimuli as input (e.g., collections of dots) under unsupervised learning. The resultant network is able to account for the distance and size effects, two ubiquitous effects in numerical cognition. Furthermore, the properties of the network units conform in detail to the characteristics of recently discovered number-selective neurons. In a second study, we simulate symbol learning by presenting symbolic and nonsymbolic input simultaneously. The same number-selective neurons learn to represent the numerical meaning of symbols. In doing so, they show properties reminiscent of the originally available number-selective neurons, but at the same time, the representational efficiency of the neurons is increased when presented with symbolic input. This finding presents a concrete proposal on the linkage between higher order numerical cognition and more primitive numerical abilities and generates specific predictions on the neural substrate of number processing.

Abstract: Clustering ensembles have emerged as a powerful method for improving both the robustness and the stability of unsupervised classification solutions. However, finding a consensus clustering from multiple partitions is a difficult problem that can be approached from graph-based, combinatorial or statistical perspectives. We offer a probabilistic model of consensus using a finite mixture of multinomial distributions in a space of clusterings. A combined partition is found as a solution to the corresponding maximum likelihood problem using the EM algorithm. The excellent scalability of this algorithm and comprehensible underlying model are particularly important for clustering of large datasets. This study compares the performance of the EM consensus algorithm with other fusion approaches for clustering ensembles. We also analyze clustering ensembles with incomplete information and the effect of missing cluster labels on the quality of overall consensus. Experimental results demonstrate the effectiveness of the proposed method on large real-world datasets. keywords: unsupervised learning, clustering ensemble, consensus function, mixture model, EM algorithm.

Abstract: The paper investigates the unsupervised learning of a model of activity for a multi-camera surveillance network that can be created from a large set of observations. This enables the learning algorithm to establish links between camera views associated with an activity. The learning algorithm operates in a correspondence-free manner, exploiting the statistical consistency of the observation data. The derived model is used to automatically determine the topography of a network of cameras and to provide a means for tracking targets across the "blind" areas of the network. A theoretical justification and experimental validation of the methods are provided.

Abstract: Developing theoretical foundations for learning is a key step towards understanding intelligence. 'Learning from examples' is a paradigm in which systems (natural or artificial) learn a functional relationship from a training set of examples. Within this paradigm, a learning algorithm is a map from the space of training sets to the hypothesis space of possible functional solutions. A central question for the theory is to determine conditions under which a learning algorithm will generalize from its finite training set to novel examples. A milestone in learning theory was a characterization of conditions on the hypothesis space that ensure generalization for the natural class of empirical risk minimization (ERM) learning algorithms that are based on minimizing the error on the training set. Here we provide conditions for generalization in terms of a precise stability property of the learning process: when the training set is perturbed by deleting one example, the learned hypothesis does not change much. This stability property stipulates conditions on the learning map rather than on the hypothesis space, subsumes the classical theory for ERM algorithms, and is applicable to more general algorithms. The surprising connection between stability and predictivity has implications for the foundations of learning theory and for the design of novel algorithms, and provides insights into problems as diverse as language learning and inverse problems in physics and engineering.

Abstract: Understanding how an animal's ability to learn relates to neural activity or is altered by lesions, different attentional states, pharmacological interventions, or genetic manipulations are central questions in neuroscience. Although learning is a dynamic process, current analyses do not use dynamic estimation methods, require many trials across many animals to establish the occurrence of learning, and provide no consensus as how best to identify when learning has occurred. We develop a state-space model paradigm to characterize learning as the probability of a correct response as a function of trial number (learning curve). We compute the learning curve and its confidence intervals using a state-space smoothing algorithm and define the learning trial as the first trial on which there is reasonable certainty (>0.95) that a subject performs better than chance for the balance of the experiment. For a range of simulated learning experiments, the smoothing algorithm estimated learning curves with smaller mean integrated squared error and identified the learning trials with greater reliability than commonly used methods. The smoothing algorithm tracked easily the rapid learning of a monkey during a single session of an association learning experiment and identified learning 2 to 4 d earlier than accepted criteria for a rat in a 47 d procedural learning experiment. Our state-space paradigm estimates learning curves for single animals, gives a precise definition of learning, and suggests a coherent statistical framework for the design and analysis of learning experiments that could reduce the number of animals and trials per animal that these studies require.

Abstract: The ability to predict a student's performance could be useful in a great number of different ways associated with university-level distance learning. Students' key demographic characteristics and their marks on a few written assignments can constitute the training set for a supervised machine learning algorithm. The learning algorithm could then be able to predict the performance of new students, thus becoming a useful tool for identifying predicted poor performers. The scope of this work is to compare some of the state of the art learning algorithms. Two experiments have been conducted with six algorithms, which were trained using data sets provided by the Hellenic Open University. Among other significant conclusions, it was found that the Naive Bayes algorithm is the most appropriate to be used for the construction of a software support tool, has more than satisfactory accuracy, its overall sensitivity is extremely satisfactory, and is the easiest algorithm to implement.

Abstract: There are two open problems when finite mixture densities are used to model multivariate data: the selection of the number of components and the initialization. In this paper, we propose an online (recursive) algorithm that estimates the parameters of the mixture and that simultaneously selects the number of components. The new algorithm starts with a large number of randomly initialized components. A prior is used as a bias for maximally structured models. A stochastic approximation recursive learning algorithm is proposed to search for the maximum a posteriori (MAP) solution and to discard the irrelevant components.

Abstract: There is growing interest in the application of machine learning techniques in bioinformatics. The supervised machine learning approach has been widely applied to bioinformatics and gained a lot of success in this research area. With this learning approach researchers first develop a large training set, which is a time-consuming and costly process. Moreover, the proportion of the positive examples and negative examples in the training set may not represent the real-world data distribution, which causes concept drift. Active learning avoids these problems. Unlike most conventional learning methods where the training set used to derive the model remains static, the classifier can actively choose the training data and the size of training set increases. We introduced an algorithm for performing active learning with support vector machine and applied the algorithm to gene expression profiles of colon cancer, lung cancer, and prostate cancer samples. We compared the classification performance of active learning with that of passive learning. The results showed that employing the active learning method can achieve high accuracy and significantly reduce the need for labeled training instances. For lung cancer classification, to achieve 96% of the total positives, only 31 labeled examples were needed in active learning whereas in passive learning 174 labeled examples were required. That meant over 82% reduction was realized by active learning. In active learning the areas under the receiver operating characteristic (ROC) curves were over 0.81, while in passive learning the areas under the ROC curves were below 0.50.

Abstract: This paper presents an unsupervised algorithm for learning a finite mixture model from multivariate data. This mixture model is based on the Dirichlet distribution, which offers high flexibility for modeling data. The proposed approach for estimating the parameters of a Dirichlet mixture is based on the maximum likelihood (ML) and Fisher scoring methods. Experimental results are presented for the following applications: estimation of artificial histograms, summarization of image databases for efficient retrieval, and human skin color modeling and its application to skin detection in multimedia databases.

Abstract: Kernel methods are widely used in statistical learning for many fields, such as protein classification and image processing. We recently extend kernel methods to intrusion detection domain by introducing a new family of kernels suitable for intrusion detection. These kernels, combined with an unsupervised learning method - one-class support vector machine, are used for anomaly detection. Our experiments show that the new anomaly detection methods are able to achieve better accuracy rates than the conventional anomaly detectors.

Abstract: We consider the general problem of learning from labeled and unlabeled data. Given a set of points, some of them are labeled, and the remaining points are unlabeled. The goal is to predict the labels of the unlabeled points. Any supervised learning algorithm can be applied to this problem, for instance, Support Vector Machines (SVMs). The problem of our interest is if we can implement a classifier which uses the unlabeled data information in some way and has higher accuracy than the classifiers which use the labeled data only. Recently we proposed a simple algorithm, which can substantially benefit from large amounts of unlabeled data and demonstrates clear superiority to supervised learning methods. Here we further investigate the algorithm using random walks and spectral graph theory, which shed light on the key steps in this algorithm.

Abstract: Object detection with a learned classifier has been applied successfully to difficult tasks such as detecting faces and pedestrians. Systems using this approach usually learn the classifier offline with manually labeled training data. We present a framework that learns the classifier online with automatically labeled data for the specific case of detecting moving objects from video. Motion information is used to automatically label training examples collected directly from the live detection task video. An online learner based on the Winnow algorithm incrementally trains a task-specific classifier with these examples. Since learning occurs online and without manual help, it can continue in parallel with detection and adapt the classifier over time. The framework is demonstrated on a person detection task for an office corridor scene. In this task, we use background subtraction to automatically label training examples. After the initial manual effort of implementing the labeling method, the framework runs by itself on the scene video stream to gradually train an accurate detector.

Abstract: We consider the problem of reinforcement learning in a controlled Markov environment with multiple objective functions of the long-term average reward type. The environment is initially unknown, and furthermore may be affected by the actions of other agents, actions that are observed but cannot be predicted beforehand. We capture this situation using a stochastic game model, where the learning agent is facing an adversary whose policy is arbitrary and unknown, and where the reward function is vector-valued. State recurrence conditions are imposed throughout. In our basic problem formulation, a desired target set is specified in the vector reward space, and the objective of the learning agent is to approach the target set, in the sense that the long-term average reward vector will belong to this set. We devise appropriate learning algorithms, that essentially use multiple reinforcement learning algorithms for the standard scalar reward problem, which are combined using the geometric insight from the theory of approachability for vector-valued stochastic games. We then address the more general and optimization-related problem, where a nested class of possible target sets is prescribed, and the goal of the learning agent is to approach the smallest possible target set (which will generally depend on the unknown system parameters). A particular case which falls into this framework is that of stochastic games with average reward constraints, and further specialization provides a reinforcement learning algorithm for constrained Markov decision processes. Some basic examples are provided to illustrate these results.

Abstract: The results from 3 category learning experiments suggest that items are better remembered when they violate a salient knowledge structure such as a rule. The more salient the knowledge structure, the stronger the memory for deviant items. The effect of learning errors on subsequent recognition appears to be mediated through the imposed knowledge structure. The recognition advantage for deviant items extends to unsupervised learning situations. Exemplar-based and hypothesis-testing models cannot account for these results. The authors propose a clustering account in which deviant items are better remembered because they are differentiated from clusters that capture regularities. The function of clusters is akin to that of schemas. Their results and analyses expose connections among research in category learning, schemas, stereotypes, and analogy.