Concept drift refers to a non stationary learning problem over time. The training and the application data often mismatch in real life problems. In this report we present a context of concept drift problem 1. We focus on the issues relevant to adaptive training set formation. We present the framework and terminology, and formulate a global picture of concept drift learners design. We start with formalizing the framework for the concept drifting data in Section 1. In Section 2 we discuss the adaptivity mechanisms of the concept drift learners. In Section 3 we overview the principle mechanisms of concept drift learners. In this chapter we give a general picture of the available algorithms and categorize them based on their properties. Section 5 discusses the related research fields and Section 5 groups and presents major concept drift applications. This report is intended to give a bird's view of concept drift research field, provide a context of the research and position it within broad spectrum of research fields and applications.

Stochastic networks are a plausible representation of the relational information among entities in dynamic systems such as living cells or social communities. While there is a rich literature in estimating a static or temporally invariant network from observation data, little has been done toward estimating time-varying networks from time series of entity attributes. In this paper we present two new machine learning methods for estimating time-varying networks, which both build on a temporally smoothed $l_1$-regularized logistic regression formalism that can be cast as a standard convex-optimization problem and solved efficiently using generic solvers scalable to large networks. We report promising results on recovering simulated time-varying networks. For real data sets, we reverse engineer the latent sequence of temporally rewiring political networks between Senators from the US Senate voting records and the latent evolving regulatory networks underlying 588 genes across the life cycle of Drosophila melanogaster from the microarray time course.

We have developed an efficient algorithm for the maximum likelihood joint tracking and association problem in a strong clutter for GMTI data. By using an iterative procedure of the dynamic logic process "from vague-to-crisp," the new tracker overcomes combinatorial complexity of tracking in highly-cluttered scenarios and results in a significant improvement in signal-to-clutter ratio.

We consider the problem of learning about and comparing the consequences of dynamic treatment strategies on the basis of observational data. We formulate this within a probabilistic decision-theoretic framework. Our approach is compared with related work by Robins and others: in particular, we show how Robins's 'G-computation' algorithm arises naturally from this decision-theoretic perspective. Careful attention is paid to the mathematical and substantive conditions required to justify the use of this formula. These conditions revolve around a property we term stability, which relates the probabilistic behaviours of observational and interventional regimes. We show how an assumption of 'sequential randomization' (or 'no unmeasured confounders'), or an alternative assumption of 'sequential irrelevance', can be used to infer stability. Probabilistic influence diagrams are used to simplify manipulations, and their power and limitations are discussed. We compare our approach with alternative formulations based on causal DAGs or potential response models. We aim to show that formulating the problem of assessing dynamic treatment strategies as a problem of decision analysis brings clarity, simplicity and generality.

Recent years have seen a growing interest in interactive narrative systems that dynamically adapt story experiences in response to users’ actions, preferences, and goals. However, relatively little empirical work has investigated runtime models of user knowledge for informing interactive narrative adaptations. User knowledge about plot scenarios, story environments, and interaction strategies is critical in a range of interactive narrative contexts, such as mystery and detective genre stories, as well as narrative scenarios for education and training. This paper proposes a dynamic Bayesian network approach for modeling user knowledge in interactive narrative environments. A preliminary version of the model has been implemented for the Crystal Island interactive narrative-centered learning environment. Results from an initial empirical evaluation suggest several future directions for the design and evaluation of user knowledge models for guiding interactive narrative generation and adaptation.

We revisit the problem of feature selection in linear discriminant analysis (LDA), that is, when features are correlated. First, we introduce a pooled centroids formulation of the multiclass LDA predictor function, in which the relative weights of Mahalanobis-transformed predictors are given by correlation-adjusted $t$-scores (cat scores). Second, for feature selection we propose thresholding cat scores by controlling false nondiscovery rates (FNDR). Third, training of the classifier is based on James--Stein shrinkage estimates of correlations and variances, where regularization parameters are chosen analytically without resampling. Overall, this results in an effective and computationally inexpensive framework for high-dimensional prediction with natural feature selection. The proposed shrinkage discriminant procedures are implemented in the R package ``sda'' available from the R repository CRAN.

In many applications, such as information retrieval or gene ranking, one is given a finite set of data of interest sharing a particular property, and wishes to find other data sharing the same property. In information retrieval, for example, the finite set can be a user query, or a set of documents known to belong to a specific category, and the goal is to scan a large database of documents to identify new documents related to the query or belonging to the same category. In gene ranking, the query is a finite list of genes known to have a given function or to be associated to a given disease, and the goal is to identify new genes sharing the same property (Aerts et al., 2006). In fact this setting is ubiquitous in many applications where identifying a data of interest is difficult or expensive, e.g., because human intervention is necessary or expensive experiments are needed, while unlabeled data can be easily collected. In such cases there is a clear opportunity to alleviate the burden and cost of interesting data identification with the help of machine learning techniques. More formally, let us assign a binary label to each possible data: positive ( 1) for data of interest, negative ( 1) for other data. Unlabeled data are data for which we do not know whether 1 they are interesting or not. Denoting X the set of data, we assume that the "query" is a finite set of data P {x

We present an exact method of greatly speeding up belief propagation (BP) for a wide variety of potential functions in pairwise MRFs and other graphical models. Specifically, our technique applies whenever the pairwise potentials have been {\em truncated} to a constant value for most pairs of states, as is commonly done in MRF models with robust potentials (such as stereo) that impose an upper bound on the penalty assigned to discontinuities; for each of the $M$ possible states in one node, only a smaller number $m$ of compatible states in a neighboring node are assigned milder penalties. The computational complexity of our method is $O(mM)$, compared with $O(M^2)$ for standard BP, and we emphasize that the method is {\em exact}, in contrast with related techniques such as pruning; moreover, the method is very simple and easy to implement. Unlike some previous work on speeding up BP, our method applies both to sum-product and max-product BP, which makes it useful in any applications where marginal probabilities are required, such as maximum likelihood estimation. We demonstrate the technique on a stereo MRF example, confirming that the technique speeds up BP without altering the solution.

This survey paper categorises, compares, and summarises from almost all published technical and review articles in automated fraud detection within the last 10 years. It defines the professional fraudster, formalises the main types and subtypes of known fraud, and presents the nature of data evidence collected within affected industries. Within the business context of mining the data to achieve higher cost savings, this research presents methods and techniques together with their problems. Compared to all related reviews on fraud detection, this survey covers much more technical articles and is the only one, to the best of our knowledge, which proposes alternative data and solutions from related domains.

The paper presents a scheme for computing lower and upper bounds on the posterior marginals in Bayesian networks with discrete variables. Its power lies in its ability to use any available scheme that bounds the probability of evidence or posterior marginals and enhance its performance in an anytime manner. The scheme uses the cutset conditioning principle to tighten existing bounding schemes and to facilitate anytime behavior, utilizing a fixed number of cutset tuples. The accuracy of the bounds improves as the number of used cutset tuples increases and so does the computation time. We demonstrate empirically the value of our scheme for bounding posterior marginals and probability of evidence using a variant of the bound propagation algorithm as a plug-in scheme.