Stormwater detention ponds play an important role in urban water management for collecting and conveying rainfall runoff from urban catchment areas to nearby streams. Their purpose is not only to avoid flooding but also to reduce stream erosion and degradation caused by the direct discharge of pollutants to the stream. We model the problem of controlling the discharge rate of water from the ponds as a partially observable hybrid Markov decision process and subsequently use Uppaal Stratego for synthesizing safe and near optimal control strategies. The generated strategies are based on noisy sensor measurements of the water height in the pond, hence the underlying system is only partially observable. We present results analyzing how sensitive the synthesized strategies are with respect to the accuracy of the measurement sensors in both offline and online settings. These types of analyses not only provide insight into the robustness of the generated strategies, but they can also be used for deciding on which measurement sensors to use, thereby balancing sensor cost and accuracy.

Traffic speed forecasting has been a very active research area as it is essential for Intelligent Transportation Systems. Although a plethora of deep learning methods have been proposed for traffic speed forecasting, the majority of them can only make point-wise prediction, which may not provide enough information for critical real-world scenarios where prediction confidence also need to be estimated, e.g., route planning for ambulances and rescue vehicles. To address this issue, we propose a novel uncertainty-aware deep learning method coined Uncertainty-Aware Temporal Graph Convolutional Network (UAT-GCN). UAT-GCN employs a Graph Convolutional Network and Gated Recurrent Unit based architecture to capture spatio-temporal dependencies. In addition, UAT-GCN consists of a specialized regressor for estimating both epistemic (model-related) and aleatoric (data-related) uncertainty. In particular, UAT-GCN utilizes Monte Carlo dropout and predictive variances to estimate epistemic and aleatoric uncertainty, respectively. In addition, we also consider the recursive dependency between predictions to further improve the forecasting performance. An extensive empirical study with real datasets offers evidence that the proposed model is capable of advancing current state-of-the-arts in terms of point-wise forecasting and quantifying prediction uncertainty with high reliability. The obtained results suggest that, compared to existing methods, the RMSE and MAE of the proposed model on the SZ-taxi dataset are reduced by$2.15\%$and$7.23\%$, respectively; the RMSE and MAE of the proposed model on the Los-loop dataset are reduced by$4.17\%$and$8.53\%$, respectively.

Current state-of-the-art techniques for metagenomic binning only utilize local features for the individual DNA sequences (contigs), neglecting additional information such as the assembly graph, in which the contigs are connected according to overlapping reads, and gene markers identified in the contigs. In this paper, we propose the use of a Variational AutoEncoder (VAE) tailored to leverage auxiliary structural information about contig relations when learning contig representations for subsequent metagenomic binning. Our method, CCVAE, improves on previous work that used VAEs for learning latent representations of the individual contigs, by constraining these representations according to the connectivity information from the assembly graph. Additionally, we incorporate into the model additional information in the form of marker genes to better differentiate contigs from different genomes. Our experiments on both simulated and real-world datasets demonstrate that CCVAE outperforms current state-of-the-art techniques, thus providing a more effective method for metagenomic binning.

MOTIVATION: Despite recent advancements in sequencing technologies and assembly methods, obtaining high-quality microbial genomes from metagenomic samples is still not a trivial task. Current metagenomic binners do not take full advantage of assembly graphs and are not optimized for long-read assemblies. Deep graph learning algorithms have been proposed in other fields to deal with complex graph data structures. The graph structure generated during the assembly process could be integrated with contig features to obtain better bins with deep learning. RESULTS: We propose GraphMB, which uses graph neural networks to incorporate the assembly graph into the binning process. We test GraphMB on long-read datasets of different complexities, and compare the performance with other binners in terms of the number of High Quality (HQ) genome bins obtained. With our approach, we were able to obtain unique bins on all real datasets, and obtain more bins on most datasets. In particular, we obtained on average 17.5% more HQ bins when compared with state-of-the-art binners and 13.7% when aggregating the results of our binner with the others. These results indicate that a deep learning model can integrate contig-specific and graph-structure information to improve metagenomic binning. AVAILABILITY AND IMPLEMENTATION: GraphMB is available from https://github.com/MicrobialDarkMatter/GraphMB. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The application of machine learning techniques in the setting of road networks holds the potential to facilitate many important intelligent transportation applications. Graph Convolutional Networks (GCNs) are neural networks that are capable of leveraging the structure of a network. However, many implicit assumptions of GCNs do not apply to road networks. We introduce theRelational Fusion Network (RFN), a novel type of Graph Convolutional Network (GCN) designed specifically for road networks. In particular, we propose methods that outperform state-of-the-art GCN architectures by up to 21–40% on two machine learning tasks in road networks. Furthermore, we show that state-of-the-art GCNs may fail to effectively leverage road network structure and may not generalize well to other road networks.

Today, advanced data analysts make use of both predictive models and optimization problem solving to build data-driven decision making applications, a combination of technologies recently termed Prescriptive Analytics (PA). Current PA applications typically have multiple layers of poorly integrated components: a relational DBMS for data storage/management, ML tools for prediction, and specialized software packages for problem modeling and optimization problem solving. This complex stack leads to inefficient, labor-intensive, and error-prone PA workflows, blocking wider adoption of PA. In this paper, we present SolveDB+ -an RDBMS for PA applications which supports all PA steps with modeling, predictive, and optimization functionalities, and integrates these in a common SQL-based framework. Major SolveDB+ novelties are 1) a powerful SQL-based approach for PA problem specification and solving, 2) an extensible in-DBMS infrastructure for prediction and optimization solvers, and 3) in-DBMS modeling and management of PA models. SolveDB+ significantly improves both PA developer productivity and performance.

Cued Recall Debriefing (CRD) is a form of retrospective think aloud approach. It involves re-immersing users to a level where emotional responses are comparable to those experienced during actual interaction with a system. To validate whether the robustness of CRD would vary with the time gap between the actual and recalled event and with the affective state preceding the recall, two empirical studies with altogether 100 participants were conducted. Specifically, participants’ emotions were measured in terms of galvanic skin response (GSR) , heart rate (HR) , and self-assessment manikin (SAM) rating when they were interacting with an email client seeded with usability problems. The same measures were taken when they viewed the videoed interactions. Two between-subject variables were ‘intervening time’ (from 0 minutes up to 24 hours) and ‘intervening affect’ (images with different valence and arousal). Advanced computational models were applied to optimise the shifting of GSR/HR waves generated at the actual interaction and recall phases, which were found to be significantly correlated. The shifting process is necessary for addressing the memory effect and is a methodological innovation. Overall, CRD proved to be a robust method that can be deployed to a broad range of HCI research and practice contexts.

In this paper, we present a method for exploratory data analysis of streaming data based on probabilistic graphical models (latent variable models). This method is illustrated by concept drift tracking, using financial client data from a European regional bank. For this particular setting, the anal yzed data spans the period from April 2007 to March 2014 and therefore starts before the beginning of the financial crisis of 2008. The implied changes in the economic climate during this period manifests itself as concept drift in the underlying data generating distribution. We explore and analyze this financial client data using a probabilistic graphical modeling framework that provides an explicit representation of concept drift as an integral part of the model. We show how learning these types of models from data provides additional insight into the hidden mechanisms governing the drift in the domain. We present an iterative approach for identifying disparate factors that jointly account for the drift in the domain. This includes a semantic characterization of one of the main influencing drift factors. Based on the experiences and results obtained from analyzing the financial data, we discuss the applicability of the framework within a more general context.

The application of machine learning techniques in the setting of road networks holds the potential to facilitate many important transportation applications. Graph Convolutional Networks (GCNs) are neural networks that are capable of leveraging the structure of a network. However, many implicit assumptions of GCNs do not apply to road networks.

Road networks are a type of spatial network, where edges may be associated with qualitative information such as road type and speed limit. Unfortunately, such information is often incomplete; for instance, OpenStreetMap only has speed limits for 13% of all Danish road segments. This is problematic for analysis tasks that rely on such information for machine learning. To enable machine learning in such circumstances, one may consider the application of network embedding methods to extract structural information from the network. However, these methods have so far mostly been used in the context of social networks, which differ significantly from road networks in terms of, e.g., node degree and level of homophily (which are key to the performance of many network embedding methods).We analyze the use of network embedding methods, specifi-cally node2vec, for learning road segment embeddings in road networks. Due to the often limited availability of information on other relevant road characteristics, the analysis focuses on leveraging the spatial network structure. Our results suggest that network embedding methods can indeed be used for deriving relevant network features (that may, e.g, be used for predicting speed limits), but that the qualities of the embeddings differ from embeddings for social networks.

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: inference in these types of models is computationally more challenging and the underlying models and updating procedures may not even support closed-form solutions. In this paper we provide an overview of the main trends and principled approaches for performing inference in hybrid Bayesian networks. The methods covered in the paper are organized and discussed according to their methodological basis. We consider how the methods have been extended and adapted to also include (hybrid) dynamic Bayesian networks, and we end with an overview of established software systems supporting inference in these types of models.

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 exponential models. Our proposal makes use of non-conjugate hierarchical priors to explicitly model temporal changes of the model parameters. We also derive a novel variational inference scheme which overcomes the use of non-conjugate priors while maintaining the computational efficiency of variational methods over conjugate models. The approach is validated on three real data sets over three latent variable models.

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 steps relate to identifying the structure of the Bayesian network using the results of the (conditional) independence tests. In this paper, we describe a new approach to parallelization of the (conditional) independence testing as experiments illustrate that this is by far the most time consuming step. The proposed parallel PC algorithm is evaluated on data sets generated at random from five different real-world Bayesian networks. The algorithm is also compared empirically with a process-based approach where each process manages a subset of the data over all the variables on the Bayesian network. The results demonstrate that significant time performance improvements are possible using both approaches.

In this paper we propose a method for scaling up filter-based feature selection in classification problems. We use the conditional mutual information as filter measure and show how the required statistics can be computed in parallel avoiding unnecessary calculations. The distribution of the calculations between the available computing units is determined based on balanced incomplete block designs, a strategy first developed within the area of statistical design of experiments. We show the scalability of our method through a series of experiments on synthetic and real-world datasets.

Unconstrained influence diagrams extend the language of influence diagrams to cope with decision problems in which the order of the decisions is unspecified. Thus, when solving an unconstrained influence diagram, we not only look for an optimal policy for each decision but also for a so-called step policy specifying the next decision given the observations made so far. However, due to the complexity of the problem, temporal constraints can force the decision maker to act before the solution algorithm has finished and, in particular, before an optimal policy for the first decision has been computed. This paper addresses this problem by proposing an anytime algorithm that at any time provides a qualified recommendation for the first decisions of the problem. The algorithm performs a heuristic-based search in a decision tree representation of the problem. We provide a framework for analyzing the performance of the algorithm, and experiments based on this framework indicate that the proposed algorithm performs significantly better under time constraints than dynamic programming.

Mixtures of polynomials (MoPs) are a nonparametric density estimation technique especially designed for hybrid Bayesian networks with continuous and discrete variables. Algorithms to learn one- and multidimensional (marginal) MoPs from data have recently been proposed. In this paper, we introduce two methods for learning MoP approximations of conditional densities from data. Both approaches are based on learning MoP approximations of the joint density and the marginal density of the conditioning variables, but they differ as to how the MoP approximation of the quotient of the two densities is found. We illustrate and study the methods using data sampled from known parametric distributions, and demonstrate their applicability by learning models based on real neuroscience data. Finally, we compare the performance of the proposed methods with an approach for learning mixtures of truncated basis functions (MoTBFs). The empirical results show that the proposed methods generally yield models that are comparable to or significantly better than those found using the MoTBF-based method.

Formal model verification has proven a powerful tool for verifying and validating the properties of a system. Central to this class of techniques is the construction of an accurate formal model for the system being investigated. Unfortunately, manual construction of such models can be a resource demanding process, and this shortcoming has motivated the development of algorithms for automatically learning system models from observed system behaviors. Recently, algorithms have been proposed for learning Markov decision process representations of reactive systems based on alternating sequences of input/output observations. While alleviating the problem of manually constructing a system model, the collection/generation of observed system behaviors can also prove demanding. Consequently we seek to minimize the amount of data required. In this paper we propose an algorithm for learning deterministic Markov decision processes from data by actively guiding the selection of input actions. The algorithm is empirically analyzed by learning system models of slot machines, and it is demonstrated that the proposed active learning procedure can significantly reduce the amount of data required to obtain accurate system models.

The influence diagram framework serves as a powerful modeling tool for symmetric decision problems with a single decision maker. However, one of the main difficulties when representing decision problems using influence diagrams is eliciting the utilities and the probabilities. This makes it desirable to be able to investigate: 1) how sensitive the solution is to variations in some utility or probability parameter, and 2) how robust the solution is to joint variations over a set of parameters. In this paper, we propose a general algorithm for performing these types of analysis.