## About

106

Publications

36,799

Reads

**How we measure 'reads'**

A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more

2,551

Citations

Citations since 2017

## Publications

Publications (106)

The use of machine learning for anomaly detection is a well-studied topic within various application domains. However, the detection problem for market surveillance remains challenging due to the lack of labelled data and the nature of anomalous behaviours, which are often contextual and spread over a sequence of anomalous instances. This paper pro...

Neural networks (NNs) have shown high predictive performance, however, with shortcomings. Firstly, the reasons behind the classifications are not fully understood. Several explanation methods have been developed, but they do not provide mechanisms for users to interact with the explanations. Explanations are social, meaning they are a transfer of k...

Recent advances in statistical inference have significantly expanded the toolbox of probabilistic modeling. Historically, probabilistic modeling has been constrained to very restricted model classes, where exact or approximate probabilistic inference is feasible. However, developments in variational inference, a general form of approximate probabil...

In many modern data analysis problems, the available data is not static but, instead, comes in a streaming fashion. Performing Bayesian inference on a data stream is challenging for several reasons. First, it requires continuous model updating and the ability to handle a posterior distribution conditioned on an unbounded data set. Secondly, the und...

There is a growing interest in applying machine learning methods on large amounts of data to solve complex problems, such as prediction of events and disturbances in the power system. This paper is a comparative study of the predictive performance of state-of-the-art supervised machine learning methods. The event prediction models are trained and v...

Prediction intervals are a machine- and human-interpretable way to represent predictive uncertainty in a regression analysis. In this paper, we present a method for generating prediction intervals along with point estimates from an ensemble of neural networks. We propose a multi-objective loss function fusing quality measures related to prediction...

Defining similarity measures is a requirement for some machine learning methods. One such method is case-based reasoning (CBR) where the similarity measure is used to retrieve the stored case or set of cases most similar to the query case. Describing a similarity measure analytically is challenging, even for domain experts working with CBR experts....

Defining similarity measures is a requirement for some machine learning methods. One such method is case-based reasoning (CBR) where the similarity measure is used to retrieve the stored case or a set of cases most similar to the query case. Describing a similarity measure analytically is challenging, even for domain experts working with CBR expert...

Utilization of measurements from on-board monitoring systems of marine vessels is a part of shipbuilding industry’s digitalization phase. The data collected can be used to verify and improve vessel’s power system design. Deployment of data-driven statistical models can enhance the knowledge about the power requirements. In this study, we describe a...

Recent advances in statistical inference have significantly expanded the toolbox of probabilistic modeling. Historically, probabilistic modeling has been constrained to (i) very restricted model classes where exact or approximate probabilistic inference were feasible, and (ii) small or medium-sized data sets which fit within the main memory of the...

Keeping the electricity production in balance with the actual demand is becoming a difficult and expensive task in spite of an involvement of experienced human operators. This is due to the increasing complexity of the electric power grid system with the intermittent renewable production as one of the contributors. A beforehand information about an...

Fashion is an area that is in constant growth. The proliferation of social media and the Web, in general, has made e-shopping, thus corresponding recommender systems, increasingly important. Fashion recommender systems is a related area that we focus on in this chapter. More specifically, we present how recommender systems are used in online fashio...

Keeping the electricity production in balance with the actual demand is becoming a difficult and expensive task in spite of an involvement of experienced human operators. This is due to the increasing complexity of the electric power grid system with the intermittent renewable production as one of the contributors. A beforehand information about an...

This work addresses the challenges related to attacks on collaborative tagging systems, which often comes in a form of malicious annotations or profile injection attacks. In particular, we study various countermeasures against two types of such attacks for social tagging systems, the Overload attack and the Piggyback attack. The countermeasure sche...

This paper addresses the important problem of discerning hateful content in social media. We propose a detection scheme that is an ensemble of Recurrent Neural Network (RNN) classifiers, and it incorporates various features associated with user- related information, such as the users’ tendency towards racism or sexism. These data are fed as input t...

Deep neural networks, including recurrent networks, have been successfully applied to human activity recognition. Unfortunately, the final representation learned by recurrent networks might encode some noise (irrelevant signal components, unimportant sensor modalities, etc.). Besides, it is difficult to interpret the recurrent networks to gain insi...

Deep neural networks, including recurrent networks, have been successfully applied to human activity recognition. Unfortunately, the final representation learned by recurrent networks might encode some noise (irrelevant signal components, unimportant sensor modalities, etc.). Besides, it is difficult to interpret the recurrent networks to gain insi...

This work addresses challenges related to attacks on social tagging systems, which often comes in a form of malicious annotations or profile injection attacks. In particular, we study various countermeasures against two types of threats for such systems, the Overload and the Piggyback attacks. The studied countermeasures include baseline classifier...

In this paper we propose a scalable importance sampling algorithm for computing Gaussian mixture posteriors in conditional linear Gaussian Bayesian networks. Our contribution is based on using a stochastic gradient ascent procedure taking as input a stream of importance sampling weights, so that a mixture of Gaussians is dynamically updated with no...

This work addresses challenges related to attacks on social tagging systems, which often comes in a form of malicious annotations or profile injection attacks. In particular, we study various countermeasures against two types of threats for such systems, the Overload and the Piggyback attacks. The studied countermeasures include baseline classifier...

Hybrid Bayesian networks have received an increasing attention during the last years. The difference with respect to standard Bayesian networks is that they can host discrete and continuous variables simultaneously, which extends the applicability of the Bayesian network framework in general. However, this extra feature also comes at a cost: infere...

This paper addresses the important problem of discerning hateful content in social media. We propose a detection scheme that is an ensemble of Recurrent Neural Network (RNN) classifiers, and it incorporates various features associated with user-related information, such as the users' tendency towards racism or sexism. These data are fed as input to...

This paper addresses the important problem of discerning hateful content in social media. We propose a detection scheme that is an ensemble of Recurrent Neural Network (RNN) classifiers, and it incorporates various features associated with user-related information, such as the users’ tendency towards racism or sexism. These data
are fed as input to...

This paper is concerned with paraphrase detection. The ability to detect similar sentences written in natural language is crucial for several applications, such as text mining, text summarization, plagiarism detection, authorship authentication and question answering. Given two sentences, the objective is to detect whether they are semantically ide...

We introduce Poisson Matrix Factorization with Content and Social trust information (PoissonMF-CS), a latent variable probabilistic model for recommender systems with the objective of jointly modeling social trust, item content and user’s preference using Poisson matrix factorization framework. This probabilistic model is equivalent to collectively...

In recent years, research has been done on applying Recurrent Neural Networks (RNNs) as recommender systems. Results have been promising, especially in the session-based setting where RNNs have been shown to outperform state-of-the-art models. In many of these experiments, the RNN could potentially improve the recommendations by utilizing informati...

Making inferences from data streams is a pervasive problem in many modern data analysis applications. But it requires to address the problem of continuous model updating and adapt to changes or drifts in the underlying data generating distribution. In this paper, we approach these problems from a Bayesian perspective covering general conjugate expo...

In recent years, research has been done on applying Recurrent Neural Networks (RNNs) as recommender systems. Results have been promising, especially in the session-based setting where RNNs have been shown to outperform state-of-the-art models. In many of these experiments, the RNN could potentially improve the recommendations by utilizing informati...

In this paper we present an approach for scaling up Bayesian learning using variational methods by exploiting distributed computing clusters managed by modern big data processing tools like Apache Spark or Apache Flink, which efficiently support iterative map-reduce operations. Our approach is defined as a distributed projected natural gradient asc...

The AMIDST Toolbox is a software for scalable probabilistic machine learning with a spe- cial focus on (massive) streaming data. The toolbox supports a flexible modeling language based on probabilistic graphical models with latent variables and temporal dependencies. The specified models can be learnt from large data sets using parallel or distribu...

In this paper, we study the maximum a posteriori (MAP) problem in dynamic hybrid Bayesian networks. We are interested in finding the sequence of values of a class variable that maximizes the posterior probability given evidence. We propose an approximate solution based on transforming the MAP problem into a simpler belief update problem. The propos...

This paper considers a parallel algorithm for Bayesian network structure learning from large data sets. The parallel algorithm is a variant of the well known PC algorithm. The PC algorithm is a constraint-based algorithm consisting of five steps where the first step is to perform a set of (conditional) independence tests while the remaining four st...

In this paper we analyse the problem of probabilistic inference in CLG networks when evidence comes in streams. In such situations, fast and scalable algorithms, able to provide accurate responses in a short time are required. We consider the instantiation of variational inference and importance sampling, two well known tools for probabilistic infe...

This paper describes a parallel version of the PC algorithm for learning the structure of a Bayesian network from data. The PC algorithm is a constraint-based algorithm consisting of five steps where the first step is to perform a set of (conditional) independence tests while the remaining four steps relate to identifying the structure of the Bayes...

An often used approach for detecting and adapting to concept drift when doing classification is to treat the data as i.i.d. and use changes in classification accuracy as an indication of concept drift. In this paper, we take a different perspective and propose a framework, based on probabilistic graphical models, that explicitly represents concept...

This paper describes some of the key properties of the pro- posed solution for the RecSys 2015 Challenge from the team Tøyvind thørrud. Three contributions will be highlighted: i) Feature extraction, ii) Classifier design, and iii) Decision rules to optimize the prediction results towards the RecSys Challenge’s score. We finished sixth out of more...

This paper studies data-driven short-term load forecasting, where historic data are used to predict the expected load for the next 24 h. Our focus is to simplify and automate the estimation and analysis of various forecasting models. We propose a three-stage approach to load forecasting, consisting of preprocessing, forecasting, and postprocessing,...

Mixtures of Truncated Basis Functions (MoTBFs) have recently been proposed for modelling univariate and joint distributions in hybrid Bayesian networks. In this paper we analyse the problem of learning conditional MoTBF distributions from data. Our approach utilizes a new technique for learning joint MoTBF densities, then propose a method for using...

Given evidence on a set of variables in a Bayesian network, the most probable explanation (MPE) is the problem of finding a configuration of the remaining variables with maximum posterior probability. This problem has previously been addressed for discrete Bayesian networks and can be solved using inference methods similar to those used for finding...

Collaborative filtering has emerged as a popular way of making user recommendations, but with the increasing sizes of the underlying databases scalability is becoming a crucial issue. In this paper we focus on a recently proposed probabilistic collaborative filtering model that explicitly represents all users and items simultaneously in the model....

Fashion e-commerce is a fast growing area in online shopping. The fashion domain has several interesting properties, which make personalised recommendations more difficult than in more traditional domains. To avoid potential bias when using explicit user ratings, which are also expensive to obtain, this work approaches fashion recommendations by an...

The framework of Bayesian networks is a widely popular formalism for performing belief update under uncertainty. Structure restricted Bayesian network models such as the Naive Bayes Model and Tree-Augmented Naive Bayes (TAN) Model have shown impressive performance for solving classification tasks. However, if the number of variables or the amount o...

Monitoring a complex process often involves keeping an eye on hundreds or thousands of sensors to determine whether or not the process is stable. We have been working with dynamic data from an oil production facility in the North sea, where unstable situations should be identified as soon as possible. Motivated by this problem setting, we propose a...

Load prediction is an important tool for grid utilities and power companies for managing the power system. This has traditionally mainly been applied at transmission and sub-transmission system levels. However, as the traditional grid evolves into a smart grid, load prediction at smaller scales becomes necessary for efficient management and operati...

The rollout of advanced metering infrastructure that is planned in many countries worldwide will lead to a massive inflow of data from moderately reliable sensory equipment. In principle, this will make intelligent and automated planning and operation possible at an increasingly finer scale in the electric grid. However, errors can creep into the m...

In this paper we look at statistical models for predicting the outcome of football matches in a league. That is, our aim is to find a statistical model which, based on the game-history so far in a season, can predict the outcome of next round's matches. Many such models exist, but we are not aware of a thorough comparison of the models' merits as b...

In this paper we describe a new method for learning hybrid Bayesian network models from data. The method utilizes a kernel density estimator, which is in turn "translated" into a mixture of truncated basis functions-representation using a convex optimization technique. We argue that these estimators approximate the maximum likelihood estimators, an...

In this paper we study the problem of exact inference in hybrid Bayesian networks using mixtures of truncated basis functions (MoTBFs). We propose a structure for handling probability potentials called Sum-Product factorized potentials, and show how these po-tentials facilitate efficient inference based on i) properties of the MoTBFs and ii) ideas...

Recommender systems based on collaborative filtering have received a great deal of interest over the last decade. Typically, these types of systems either take a user-centered or an item-centered approach when making recommendations, but by employing only one of these two perspectives we may unintentionally leave out important information that coul...

In this paper we propose a framework, called mixtures of truncated basis functions (MoTBFs), for representing general hybrid Bayesian networks. The proposed framework generalizes both the mixture of truncated exponentials (MTEs) framework and the Mixture of Polynomials (MoPs) framework. Similar to MTEs and MoPs, MoTBFs are defined so that the poten...

The overall goal of this research is to study reasoning under uncertainty by combining Bayesian Networks and Case-Based Reasoning through constructing an experimental decision support system for classification of cancer pain. We have experimentally analysed a medical dataset in order to reveal properties of the data with respect to properties of th...

In complex domains, it is often necessary to determine which reasoning method would be the most appropriate for each task, and a combination of different methods has often shown the best results. We examine how two complementary reasoning methods, case-based reasoning and Bayesian networks, can be combined using metareasoning to form a more robust...

In this paper we discuss different architectures for reasoning under uncertainty related to our ongoing research into building
a medical decision support system. The uncertainty in the medical domain can be divided into a well understood part and a
less understood part. This motivates the use of a hybrid decision support system, and in particular,...

Bayesian Networks are useful for solving a wide range of problems in many domains. Yet, they are exposed to one important challenge when structural and parametrical changes occur. As Bayesian networks lack memory regarding changes over time, there is currently no good way of maintaining a history of changes and their provenance. Thus, any variance...

Bayesian networks with mixtures of truncated exponentials (MTEs) support efficient inference algorithms and provide a flexible way of modeling hybrid domains (domains containing both discrete and continuous variables). On the other hand, estimating an MTE from data has turned out to be a difficult task, and most prevalent learning methods treat par...

Monitoring a complex process often involves keeping an eye on hundreds or thousands of sensors to determine whether or not the process is in a normal state. We have been working with data from an oil production facility in the North

Naive Bayes (NB) models are among the simplest probabilistic classifiers. However, they often perform surprisingly well in practice, even though they are based on the strong assumption that all attributes are conditionally independent given the class variable. The vast majority of research on NB models assume that the conditional probability tables...

Bayesian networks with mixtures of truncated exponentials (MTEs) are gaining popularity as a flexible modelling framework for hybrid domains. MTEs support efficient and exact inference algorithms, but estimating an MTE from data has turned out to be a difficult task. Current methods suffer from a considerable computational burden as well as the ina...

One of the simplest, and yet most consistently well-performing set of classifiers is the naïve Bayes models (a special class of Bayesian network models). However, these models rely on the (naïve) assumption that all the attributes used to describe an instance are conditionally independent given the class of that instance. To relax this independence...

Since the 1980s, Bayesian networks (BNs) have become increasingly popular for building statistical models of complex systems. This is particularly true for boolean systems, where BNs often prove to be a more efficient modelling framework than traditional reliability techniques (like fault trees and reliability block diagrams). However, limitations...

We describe a procedure for inducing conditional densities within the mixtures of truncated exponentials (MTE) framework. We analyse possible conditional MTE specifications and propose a model selection scheme, based on the BIC score, for partitioning the domain of the conditioning variables. Finally, experimental results demonstrate the applicabil...

This paper discusses the traditional cost-benefit analysis in the context of prioritization of risk reducing measures. Such measures are often uncertain both in terms of cost of the measures and their effects. We introduce the concept of decision flexibility to develop an alternative framework for making such decisions.

Bayesian network (BN) models gain more and more popularity as a tool in reliability analysis. In this paper we consider some of the properties of BNs that have made them popular, consider some of the recent developments, and also point to the most important remaining challenges when using BNs in reliability. 1. The Good: The foundation of Bayesian...

Bayesian networks with mixtures of truncated exponentials (MTEs) support efficient in-ference algorithms and provide a flexible way of modeling hybrid domains. On the other hand, estimating an MTE from data has turned out to be a difficult task, and most preva-lent learning methods treat parameter estimation as a regression problem. The drawback of...

Over the last decade, Bayesian networks (BNs) have become a popular tool for modelling many kinds of statistical problems. We have also seen a growing interest for using BNs in the reliability analysis community. In this paper we will discuss the properties of the modelling framework that make BNs particularly well suited for reliability applicatio...

Over the last decade, Bayesian Networks (BNs) have become a popular tool for modelling many kinds of statistical problems. We have also seen a growing interest for using BNs in the reliability analysis community. This article discusses the properties of the modelling framework that are of highest importance for reliability practitioners.

Over the last decade, Bayesian networks (BNs) have become a popular tool for modeling many kinds of statistical problems. In this chapter we will discuss the properties of the modeling framework that make BNs particularly well suited for reliability applications. This discussion is closely linked to the analysis of a real-world example.

Over the last decade, Bayesian networks (BNs) have become a popular tool for modeling many kinds of statistical problems. In this chapter we will discuss the properties of the modeling framework that make BNs particularly well suited for reliability applications. This discussion is closely linked to the analysis of a real-world example.

Classification problems have a long history in the machine learning literature. One of the simplest, and yet most consistently
well-performing set of classifiers is the Naïve Bayes models. However, an inherent problem with these classifiers is the assumption
that all attributes used to describe an instance are conditionally independent given the cl...