Search

This article proposes a novel study on personality recognition using video data from different scenarios. Our goal is to jointly model nonverbal behavioral cues with contextual information for a robust, multi-scenario, personality recognition system. Therefore, we build a novel multi-stream Convolutional Neural Network (CNN) framework, which considers multiple sources of information. From a given scenario, we extract spatio-temporal motion descriptors from every individual in the scene, spatio-temporal motion descriptors encoding social group dynamics, and proxemics descriptors to encode the interaction with the surrounding context. All the proposed descriptors are mapped to the same feature space facilitating the overall learning effort. Experiments on two public datasets demonstrate the effectiveness of jointly modeling the mutual Person-Context information, outperforming the state-of-the art-results for personality recognition in two different scenarios. Last, we present CNN class activation maps for each personality trait, shedding light on behavioral patterns linked with personality attributes.

In this paper we propose two methods for instance transfer based on conformal prediction. As a distinctive character, both of the methods are model independent and combine feature selection and source-instance selection to avoid negative transfer. The methods have been tested experimentally for different types of classification model on several benchmark data sets. The experimental results demonstrate that the new methods are capable of outperforming significantly standard instance transfer methods. (C) 2019 Elsevier B.V. All rights reserved.

This report summarises the Digital Ludeme Project, a recently launched 5-year research project being conducted at Maastricht University. This computational study of the world's traditional strategy games seeks to improve our understanding of early games, their development, and their role in the spread of related mathematical ideas throughout recorded human history.

Background/Objective Current severe traumatic brain injury (TBI) outcome prediction models calculate the chance of unfavourable outcome after 6 months based on parameters measured at admission. We aimed to improve current models with the addition of continuously measured neuromonitoring data within the first 24 h after intensive care unit neuromonitoring. Methods Forty-five severe TBI patients with intracranial pressure/cerebral perfusion pressure monitoring from two teaching hospitals covering the period May 2012 to January 2019 were analysed. Fourteen high-frequency physiological parameters were selected over multiple time periods after the start of neuromonitoring (0–6 h, 0–12 h, 0–18 h, 0–24 h). Besides systemic physiological parameters and extended Corticosteroid Randomisation after Significant Head Injury (CRASH) score, we added estimates of (dynamic) cerebral volume, cerebral compliance and cerebrovascular pressure reactivity indices to the model. A logistic regression model was trained for each time period on selected parameters to predict outcome after 6 months. The parameters were selected using forward feature selection. Each model was validated by leave-one-out cross-validation. Results A logistic regression model using CRASH as the sole parameter resulted in an area under the curve (AUC) of 0.76. For each time period, an increased AUC was found using up to 5 additional parameters. The highest AUC (0.90) was found for the 0–6 h period using 5 parameters that describe mean arterial blood pressure and physiological cerebral indices. Conclusions Current TBI outcome prediction models can be improved by the addition of neuromonitoring bedside parameters measured continuously within the first 24 h after the start of neuromonitoring. As these factors might be modifiable by treatment during the admission, testing in a larger (multicenter) data set is warranted.

In this paper, we introduce an unsupervised hierarchical framework for modeling trajectories in surveillance scenarios. Inspired by the object recognition field, a novel feature representation optimized for a neural network learning architecture is proposed. Low levels of the hierarchy capture local spatio-temporal motion attributes such as spatial orientation and speed, while higher levels contribute to obtaining richer semantic information. The bottom-up construction of the hierarchical framework exploits the inherent statistical correlations between neighboring elements using an increasing spatio-temporal grid. Cross-entropy based optimization in combination with autoencoders is used to learn weights for subsequent hierarchical layers. Finally, the Bayesian probabilistic framework built on top of the hierarchical model is proposed for applications such as long-term path prediction and abnormality detection. We demonstrate the efficiency of the proposed model on both indoor and outdoor datasets, achieving results comparable with state-of-the-art methods. (C) 2019 Elsevier B.V. All rights reserved.

People express their emotions through multiple channels, such as visual and audio ones. Consequently, automatic emotion recognition can be significantly benefited by multimodal learning. Even-though each modality exhibits unique characteristics; multimodal learning takes advantage of the complementary information of diverse modalities when measuring the same instance, resulting in enhanced understanding of emotions. Yet, their dependencies and relations are not fully exploited in audio–video emotion recognition. Furthermore, learning an effective metric through multimodality is a crucial goal for many applications in machine learning. Therefore, in this article, we propose multimodal emotion recognition metric learning (MERML), learned jointly to obtain a discriminative score and a robust representation in a latent-space for both modalities. The learned metric is efficiently used through the radial basis function (RBF) based support vector machine (SVM) kernel. The evaluation of our framework shows a significant performance, improving the state-of-the-art results on the eNTERFACE and CREMA-D datasets.

Imagine computer programs (agents) that learn to coordinate or to compete. This study investigates how their learning processes influence each other. Such adaptive agents already take vital roles behind the scenes of our society, e.g., high frequency automated traders participate in financial trading and create more volume than human trading in some US markets. However, many learning algorithms only have proven performance guarantees if they act alone - as soon as a second agent influences the outcomes most guarantees are invalid. This dissertation extends guarantees to strategic interactions of several agents and examines how closely algorithms approximate optimal behavior.This research was funded by a TopTalent 2008 grant of NWO.

The accuracy and reliability of weather forecasting are of importance for many economic, business and management activities. This paper introduces novel data-driven predictive models based on deep convolutional neural networks (CNN) architecture for temperature and wind speed prediction in weather data. In particular, the proposed deep learning framework employs different upgrading versions of the convolutional neural networks i.e. 1d-, 2d- and 3d-CNN. The introduced models exploit the spatio-temporal multivariate weather data for learning shared representations using historical data and forecasting weather elements for a number of user defined weather stations simultaneously in an end-to-end fashion. The embedded feature learning component of the models as well as coupling the learned features of different input layers have shown to have a significant impact on the prediction task. The proposed models show promising results compared to the classical neural networks architecture used for modeling nonlinear systems. Two experimental setups have been considered based on a dataset collected from the Weather Underground website at six stations located in Netherlands and Belgium as well as a larger dataset with higher temporal resolution from the National Climatic Data Center (NCDC) at five stations located in Denmark. First, we focus on simultaneously predicting the temperature of two main stations of Amsterdam and Brussels for 1-10 days ahead. The second experiment concerns wind speed prediction at three weather stations located in Denmark for 6 and 12 h ahead. The obtained numerical results show that learning new shared representations of the weather data by means of convolutional operations improves the prediction performance. (C) 2019 Elsevier B.V. All rights reserved.

Social complexity requires people to create new ways of interacting with one another to counteract new social boundaries. Board games provide one avenue by which archaeologists may examine the ways in which people interacted. The act of play is liminoid, allowing for people to more easily overcome social boundaries to interaction, while also creating a shared experience for the process of grounding. Thus, games provide an opportunity for people to build relationships and also to negotiate their identity. This research counters previous assertions that only complex societies play strategy games. In Bronze Age Cyprus, middle-range society became increasingly complex with the development of urbanism and differential access to wealth, while also producing a particularly large corpus of ancient board games. Determining where games were played, and then analyzing the social context of games in those spaces using multiple correspondence analysis, it was found that people were more likely to be playing in public during periods of greater complexity. This is likely due to the performative aspects of play, allowing for the communication of status in different ways when games are played in public spaces. The pattern found shows the variation in the ways people incorporated games into their social strategies during periods of different types of complexity.

BACKGROUND: The present study examined food cravings in daily life by comparing overweight and normal-weight participants right before eating events and at non-eating moments. It was hypothesised that overweight participants would have (i) more frequent, (ii) stronger and (iii) a greater variety of high-caloric palatable food cravings, and also would (iv) consume more high-caloric palatable foods, than normal-weight participants.METHODS: Ecological momentary assessment (EMA) was used to assess food craving strength and frequency, variety of specific food cravings, and food intake. Fifty-seven overweight and 43 normal-weight adult participants were assessed at eating events and at an average of eight random non-eating moments per day for 2 weeks. Foods were categorised as: high-caloric high palatable foods (HCHP), fruits and salads, staple food dishes and sandwiches, and soups and yoghurts.RESULTS: Overweight participants reported more frequent HCHP food cravings specifically at non-eating moments than did normal-weight participants. Normal-weight participants reported more food cravings for staple foods, specifically at eating events. Moreover, overweight participants craved a greater variety of HCHP foods than normal-weight participants at both eating events and random non-eating moments. No other significant between-group differences were found.CONCLUSIONS: The results highlight the importance for obesity interventions (i) to specifically target high-caloric palatable food cravings that are experienced during the day and are not tied to eating moments and (ii) to aim for a reduction in the variety of high-caloric palatable food cravings. It might be fruitful to deliver treatment aimed at reducing cravings via mobile devices because this allows for easy individual tailoring and timing of interventions.

One approach to game design is to start with existing ideas and to modify them in interesting new ways. This process of “creative plagiarism” is applied to the Japanese logic puzzle Sudoku to create a new variant called Ludoku, which simplifies some aspects of the original puzzle while adding new strategies without adding undue complexity. This paper describes the design process from initial conception to final product, and how computer analysis was crucial to its success.

The COVID-19 pandemic forces many robotics teachers to rethink their approach to education. Distance rules and the constant threat of a partial or complete lockdown leading to limited access to classroom equipment make it challenging to plan for hands-on education where students experience robotics by experimenting and studying with robotic hardware. On the other hand, this hands-on active learning experience is one of the strengths of robotics education and the ability to handle hardware equipment a substantial learning goal of study programs on robotics. In this paper we present and discuss the approach taken for the course Robotics and Embedded Systems at Maastricht University. The course had been adjusted to meet COVID-19 safety regulations and to allow for a fast seamless transition between onsite education at university and online education where students can work with robotic hardware at home. We share experience, best practice advice as well as educational material to help other teachers benefit from our developments. A key contribution is our custom-made, low-cost, educational modular robotic system for teaching kinematics, locomotion, and PID control that we make publicly available for replication through the website https://www.maastrichtuniversity.nl/edmo.

Harmony in visual compositions is a concept that cannot be defined or easily expressed mathematically, even by humans. The goal of the research described in this paper was to find a numerical representation of artistic compositions with different levels of harmony. We ask humans to rate a collection of grayscale images based on the harmony they convey. To represent the images, a set of special features were designed and extracted. By doing so, it became possible to assign objective measures to subjectively judged compositions. Given the ratings and the extracted features, we utilized machine learning algorithms to evaluate the efficiency of such representations in a harmony classification problem. The best performing model (SVM) achieved 80% accuracy in distinguishing between harmonic and disharmonic images, which reinforces the assumption that concept of harmony can be expressed in a mathematical way that can be assessed by humans., Comment: 9 pages, ICAART 2021

Background Flow and mass cytometry are important modern immunology tools for measuring expression levels of multiple proteins on single cells. The goal is to better understand the mechanisms of responses on a single cell basis by studying differential expression of proteins. Most current data analysis tools compare expressions across many computationally discovered cell types. Our goal is to focus on just one cell type. Our narrower field of application allows us to define a more specific statistical model with easier to control statistical guarantees. Results Differential analysis of marker expressions can be difficult due to marker correlations and inter-subject heterogeneity, particularly for studies of human immunology. We address these challenges with two multiple regression strategies: a bootstrapped generalized linear model and a generalized linear mixed model. On simulated datasets, we compare the robustness towards marker correlations and heterogeneity of both strategies. For paired experiments, we find that both strategies maintain the target false discovery rate under medium correlations and that mixed models are statistically more powerful under the correct model specification. For unpaired experiments, our results indicate that much larger patient sample sizes are required to detect differences. We illustrate the R package and workflow for both strategies on a pregnancy dataset. Conclusion Our approach to finding differential proteins in flow and mass cytometry data reduces biases arising from marker correlations and safeguards against false discoveries induced by patient heterogeneity.

Introduction Electrocardiographic imaging (ECGI) has been used to investigate arrhythmia mechanisms in various conditions. Data on normal human subjects, especially in Europe, are scarce. Dynamic characteristics of ventricular activation and recovery during sinus rhythm have not been assessed before. Purpose To examine cardiac electrophysiology and its dynamic aspects in normal subjects using ECGI, in order to provide a range of normal patterns and values for activation (AT) and recovery times (RT), activation-recovery intervals (ARI, a surrogate for action potential duration) and their dynamicity. Methods 11 Subjects (age 57±7 years, 27% male, all normal LVEF) with atypical chest pain who underwent a cardiac CT-scan as part of clinical care but who were negative for any pathology on full examination were included. A validated non-commercial potential-based formulation of ECGI was used to reconstruct unipolar electrograms (EGMs) on the epicardial surface for three sinus beats within minutes from each other, per individual. ATs and RTs were determined as the maximum negative upslope during QRS, and maximum positive upslope during T wave of the local EGMs. Additionally, we determined locations of first and last activation and recovery. Inter- and intra-individual differences were computed. Results Subjects had normal 12-lead characteristics without ST-deviations, and an average QTc interval of 415±18ms. Figure 1, panel A shows ECGI during sinus rhythm for 3 representative subjects, and panel B summarizes all findings on the entire epicardium. The first epicardial activation breakthrough typically occurred on the right ventricle (RV), consistent with the concept that the thinner RV wall accounts for a faster transmural activation. Last activation was mostly on the base of the left ventricle (LV), on the inferior to lateral wall. Earliest recovery occurred predominantly on the anterior surface, while latest recovery occurred on the inferior surface. Complete activation of the epicardial surface (from earliest to latest AT) took 41±8ms, while recovery (earliest AT to latest RT) took 317±24ms and average ARI (local AT to local RT) took 232±23ms. Thus, inter-individual variation of recovery duration was higher than of activation. Intra-individual differences between beats in ATs, RTs and ARIs of distinct sinus beats were small (2.3±3.1ms, 9.7±8.8ms and 9.8±9.1ms, respectively) suggesting that ECGI enables stable reconstruction quality (Figure 1, panel C). Conclusion In this cohort, noninvasive ECGI provides novel insights in ventricular electrophysiology. Electrical recovery is more variable than activation, both intra-individually and inter-individually. Overall, AT, RT and ARI differences between sinus beats were low. ECGI appears suitable to assess dynamic electrical patterns during cardiac pathology. Figure 1 Funding Acknowledgement Type of funding source: None

The growth of social media has revolutionized the way people access information. Although platforms like Facebook and Twitter allow for a quicker, wider and less restricted access to information, they also consist of a breeding ground for the dissemination of fake news. Most of the existing literature on fake news detection on social media proposes user-based or content-based approaches. However, recent research revealed that real and fake news also propagate significantly differently on Twitter. Nonetheless, only a few articles so far have explored the use of propagation features in their detection. Additionally, most of them have based their analysis on a narrow tweet retrieval methodology that only considers tweets to be propagating a news piece if they explicitly contain an URL link to an online news article. By basing our analysis on a broader tweet retrieval methodology that also allows tweets without an URL link to be considered as propagating a news piece, we contribute to fill this research gap and further confirm the potential of using propagation features to detect fake news on Twitter. We firstly show that real news are significantly bigger in size, are spread by users with more followers and less followings, and are actively spread on Twitter for a longer period of time than fake news. Secondly, we achieve an 87% accuracy using a Random Forest Classifier solely trained on propagation features. Lastly, we design a Geometric Deep Learning approach to the problem by building a graph neural network that directly learns on the propagation graphs and achieve an accuracy of 73.3%.

Highly exposed seronegative (HESN) individuals present a unique setting to study mechanisms of protection against HIV acquisition. As natural killer (NK) cell activation and function have been implicated as a correlate of protection in HESN individuals, we sought to better understand the features of NK cells that may confer protection. We used mass cytometry to phenotypically profile NK cells from a cohort of Beninese sex workers and healthy controls.We found that NK cells from HESN women had increased expression of NKG2A, NKp30 and LILRB1, as well as the Fc receptor CD16, and decreased expression of DNAM-1, CD94, Siglec-7, and NKp44. Using functional assessments of NK cells from healthy donors against autologous HIV-infected CD4+ T cells, we observed that NKp30+ and Siglec-7+ cells had improved functional activity. Further, we found that NK cells from HESN women trended towards increased antibody-dependent cellular cytotoxicity (ADCC) activity; this activity correlated with increased CD16 expression. Overall, we identify features of NK cells in HESN women that may contribute to protection from HIV infection. Follow up studies with largercohorts are warranted to confirm these findings.

Accessing large, manually annotated audio databases in an effort to create robust models for emotion recognition is a notably difficult task, handicapped by the annotation cost and label ambiguities. On the contrary, there are plenty of publicly available datasets for emotion recognition which are based on facial expressivity due to the prevailing role of computer vision in deep learning research, nowadays. Thereby, in the current work, we performed a study on cross-modal transfer knowledge between audio and facial modalities within the emotional context. More concretely, we investigated whether facial information from videos could be used to boost the awareness and the prediction tracking of emotions in audio signals. Our approach was based on a simple hypothesis: that the emotional state's content of a person's oral expression correlates with the corresponding facial expressions. Research in the domain of cognitive psychology was affirmative to our hypothesis and suggests that visual information related to emotions fused with the auditory signal is used from humans in a cross-modal integration schema to better understand emotions. In this regard, a method called dacssGAN (which stands for Domain Adaptation Conditional Semi-Supervised Generative Adversarial Networks) is introduced in this work, in an effort to bridge these two inherently different domains. Given as input the source domain (visual data) and some conditional information that is based on inductive conformal prediction, the proposed architecture generates data distributions that are as close as possible to the target domain (audio data). Through experimentation, it is shown that classification performance of an expanded dataset using real audio enhanced with generated samples produced using dacssGAN (50.29% and 48.65%) outperforms the one obtained merely using real audio samples (49.34% and 46.90%) for two publicly available audio-visual emotion datasets. (C) 2019 The Authors. Published by Elsevier B.V.

This paper introduces an information-theoretic method for selecting a subset of problems which gives the most information about a group of problem-solving algorithms. This method was tested on the games in the General Video Game AI (GVGAI) framework, allowing us to identify a smaller set of games that still gives a large amount of information about the abilities of different game-playing agents. This approach can be used to make agent testing more efficient. We can achieve almost as good discriminatory accuracy when testing on only a handful of games as when testing on more than a hundred games, something which is often computationally infeasible. Furthermore, this method can be extended to study the dimensions of the effective variance in game design between these games, allowing us to identify which games differentiate between agents in the most complementary ways., Comment: 8 pages, 1 figure, 2 tables

The physics-based simulation game Angry Birds has been heavily researched by the AI community over the past five years, and has been the subject of a popular AI competition that is currently held annually as part of a leading AI conference. Developing intelligent agents that can play this game effectively has been an incredibly complex and challenging problem for traditional AI techniques to solve, even though the game is simple enough that any human player could learn and master it within a short time. In this paper we analyse how hard the problem really is, presenting several proofs for the computational complexity of Angry Birds. By using a combination of several gadgets within this game's environment, we are able to demonstrate that the decision problem of solving general levels for different versions of Angry Birds is either NP-hard, PSPACE-hard, PSPACE-complete or EXPTIME-hard. Proof of NP-hardness is by reduction from 3-SAT, whilst proof of PSPACE-hardness is by reduction from True Quantified Boolean Formula (TQBF). Proof of EXPTIME-hardness is by reduction from G2, a known EXPTIME-complete problem similar to that used for many previous games such as Chess, Go and Checkers. To the best of our knowledge, this is the first time that a single-player game has been proven EXPTIME-hard. This is achieved by using stochastic game engine dynamics to effectively model the real world, or in our case the physics simulator, as the opponent against which we are playing. These proofs can also be extended to other physics-based games with similar mechanics., Comment: 55 Pages, 39 Figures

Specific causes of preterm birth remain unclear. Several recent studies have suggested that immune changes during pregnancy are associated with the timing of delivery, yet few studies have been performed in low-income country settings where the rates of preterm birth are the highest. We conducted a retrospective nested case-control evaluation within a longitudinal study among HIV-uninfected pregnant Kenyan women. To characterize immune function in these women, we evaluated unstimulated and stimulated peripheral blood mononuclear cells in vitro with the A/California/2009 strain of influenza to understand the influenza-induced immune response. We then evaluated transcript expression profiles using the Affymetrix Human GeneChip Transcriptome Array 2.0. Transcriptional profiles of sufficient quality for analysis were obtained from 54 women; 19 of these women delivered 37 weeks. The median time to birth from sample collection was 13 weeks. No transcripts were significantly associated with preterm birth in a case-control study of matched term and preterm birth (n = 42 women). In the influenza-stimulated samples, expression of IFNL1 was associated with longer time to delivery-the amount of time between sample collection and delivery (n = 54 women). A qPCR analysis confirmed that influenza-induced IFNL expression was associated with longer time to delivery. These data indicate that during pregnancy, ex vivo influenza stimulation results in altered transcriptional response and is associated with time to delivery in cohort of women residing in an area with high preterm birth prevalence.

The tiled convolutional neural network (TCNN) has been applied only to computer vision for learning invariances. We adjust its architecture to NLP to improve the extraction of the most salient features for sentiment analysis. Knowing that the major drawback of the TCNN in the NLP field is its inflexible filter structure, we propose a novel architecture called hybrid tiled convolutional neural network (HTCNN) that applies a filter only on the words that appear in similar contexts and on their neighbouring words (a necessary step for preventing the loss of some n-grams). The experiments on the IMDB movie reviews dataset demonstrate the effectiveness of the HTCNN that has a higher level of performance of more than 3% and 1% respectively than both the convolutional neural network (CNN) and the TCNN. These results are confirmed by the SemEval-2017 dataset where the recall of the HTCNN model exceeds by more than six percentage points the recall of its simple variant, CNN.

In clinical trials, animal and cell line models are often used to evaluate the potential toxic effects of a novel compound or candidate drug before progressing to human trials. However, relating the results of animal and in vitro model exposures to relevant clinical outcomes in the human in vivo system still proves challenging, relying on often putative orthologs. In recent years, multiple studies have demonstrated that the repeated dose rodent bioassay, the current gold standard in the field, lacks sufficient sensitivity and specificity in predicting toxic effects of pharmaceuticals in humans. In this study, we evaluate the potential of deep learning techniques to translate the pattern of gene expression measured following an exposure in rodents to humans, circumventing the current reliance on orthologs, and also from in vitro to in vivo experimental designs. Of the applied deep learning architectures applied in this study the convolutional neural network (CNN) and a deep artificial neural network with bottleneck architecture significantly outperform classical machine learning techniques in predicting the time series of gene expression in primary human hepatocytes given a measured time series of gene expression from primary rat hepatocytes following exposure in vitro to a previously unseen compound across multiple toxicologically relevant gene sets. With a reduction in average mean absolute error across 76 genes that have been shown to be predictive for identifying carcinogenicity from 0.0172 for a random regression forest to 0.0166 for the CNN model (p < 0.05). These deep learning architecture also perform well when applied to predict time series of in vivo gene expression given measured time series of in vitro gene expression for rats.

The “war on cancer” began over 40 years ago with the signing of the National Cancer Act of 1971. Currently, complete eradication has proven possible in early stage premetastatic disease with increasingly successful early detection and surgery protocols; however, late stage metastatic disease remains invariably fatal. One of the main causes of treatment failure in metastatic disease is the ability of cancer cells to evolve resistance to currently available therapies. Evolution of resistance to control measures is a universal problem. While it may seem that the mechanisms of resistance employed by cancer cells are impossible to control, we show that many of the resistance mechanisms are mirrored in agricultural pests. In this way, we argue that measures developed in the agricultural industry to slow or prevent pesticide resistance could be adopted in clinical cancer biology to do the same. The agriculture industry recognized the problem of pesticide resistance and responded by developing and enforcing guidelines on resistance management and prevention. These guidelines, known as integrated pest management (IPM), do not encourage eradication of pests but instead strive to maintain pests, even with the presence of resistant strains, at a level that does not cause economic damage to the crops. Integrated pest management inspired management of metastatic cancer could result in the slowing or curtailing of widespread resistance to treatment, reducing overall drug usage, and increasing the survival and quality of life of patients with cancer. Using IPM principles as a foundation and shifting the goal of treatment of metastatic disease to long-term management will require close monitoring of evolving tumor populations, judicious application of currently available therapies, and development of new criteria of success.

In the absence of curative therapies, treatment of metastatic castrate-resistant prostate cancer (mCRPC) using currently available drugs can be improved by integrating evolutionary principles that govern proliferation of resistant subpopulations into current treatment protocols. Here we develop what is coined as an ‘evolutionary stable therapy’, within the context of the mathematical model that has been used to inform the first adaptive therapy clinical trial of mCRPC. The objective of this therapy is to maintain a stable polymorphic tumor heterogeneity of sensitive and resistant cells to therapy in order to prolong treatment efficacy and progression free survival. Optimal control analysis shows that an increasing dose titration protocol, a very common clinical dosing process, can achieve tumor stabilization for a wide range of potential initial tumor compositions and volumes. Furthermore, larger tumor volumes may counter intuitively be more likely to be stabilized if sensitive cells dominate the tumor composition at time of initial treatment, suggesting a delay of initial treatment could prove beneficial. While it remains uncertain if metastatic disease in humans has the properties that allow it to be truly stabilized, the benefits of a dose titration protocol warrant additional pre-clinical and clinical investigations.

The goal of this work is to investigate the potential of making use of simple activity and motion patterns in a smart environment for approximating personality cues via machine learning techniques. Towards this goal, we present a novel framework for personality recognition, inspired by both computer vision and psychology. Results show a correlation between several behavioral features and personality traits, as well as insights of which type of everyday tasks induce stronger personality display. We experiment with the use of support vector machines, random forests and gaussian process classification achieving promising predictive ability, related to personality traits. The obtained results show consistency to a good degree, opening the path for applications in psychology, game industry, ambient assisted living, and other fields.keywordspersonality recognitionbehavior analysismachine learningpersonality traits.

Machine learning is quickly becoming an important tool in modern materials design. Where many of its successes are rooted in hugedatasets, the most common applications in academic and industrial materials design deal with datasets of at best a few tens of data points.Harnessing the power of machine learning in this context is, therefore, of considerable importance. In this work, we investigate theintricacies introduced by these small datasets. We show that individual data points introduce a significant chance factor in both modeltraining and quality measurement. This chance factor can be mitigated by the introduction of an ensemble-averaged model. This modelpresents the highest accuracy, while at the same time, it is robust with regard to changing the dataset size. Furthermore, as only a singlemodel instance needs to be stored and evaluated, it provides a highly efficient model for prediction purposes, ideally suited for the practicalmaterials scientist.

The field of robotics embraces the interconnection of a variety of disciplines not only in research but also in teaching. For students this interconnection of disciplines can provide a different quality of understanding of theory and practice since it allows breaching, recapitulating, and interlinking knowledge and skills that otherwise might be taught only separately in a variety of well-contained courses. This paper describes the approach and material of a successfully running robotics course at Maastricht University that provides students with the opportunity to study and develop locomotion control with central pattern generators on a custom-made, low-cost, versatile EDucational MOdular robotic platform called EDMO. For these studies, students follow the chain of research and development from neuroscience over mathematical modeling, control theory, programming embedded systems, to the experimentation with robotic hardware that is supported by approaches from optimization and machine learning for optimal control parameter identification. The course thus interconnects a variety of disciplines and provides students with insights for instance into locomotion control and learning and provides a vivid view on aspects of numerical mathematics and calculus. We share teaching material and hardware design files for this course that is highly appreciated by our students in the hope that other teachers and students can benefit.

Background The purpose of this study was to illustrate the additive value of computed tomography angiography (CTA) for visualisation of the coronary venous anatomy prior to cardiac resynchronisation therapy (CRT) implantation. Methods Eighteen patients planned for CRT implantation were prospectively included. A specific CTA protocol designed for visualisation of the coronary veins was carried out on a third-generation dual-source CT platform. Coronary veins were semi-automatically segmented to construct a 3D model. CTA-derived coronary venous anatomy was compared with intra-procedural fluoroscopic angiography (FA) in right and left anterior oblique views. Results Coronary venous CTA was successfully performed in all 18 patients. CRT implantation and FA were performed in 15 patients. A total of 62 veins were visualised; the number of veins per patient was 3.8 (range: 2–5). Eighty-five per cent (53/62) of the veins were visualised on both CTA and FA, while 10% (6/62) were visualised on CTA only, and 5% (3/62) on FA only. Twenty-two veins were present on the lateral or inferolateral wall; of these, 95% (21/22) were visualised by CTA. A left-sided implantation was performed in 13 patients, while a right-sided implantation was performed in the remaining 2 patients because of a persistent left-sided superior vena cava with no left innominate vein on CTA. Conclusion Imaging of the coronary veins by CTA using a designated protocol is technically feasible and facilitates the CRT implantation approach, potentially improving the outcome. Electronic supplementary material The online version of this article (10.1007/s12471-018-1132-2) contains supplementary material, which is available to authorized users.

This paper presents a novel approach to design orthogonal wavelets matched to a signal with compact support and vanishing moments. It provides a systematic and versatile framework for matching an orthogonal wavelet to a specific signal or application. The central idea is to select a wavelet by optimizing a criterion which promotes sparsity of the wavelet representation of a prototype signal. Optimization is performed over the space of orthogonal wavelet functions with compact support, coming from filter banks with a given order. Parametrizations of this space are accompanied by explicit conditions for extra vanishing moments, to be used for constrained optimization. The approach is developed first for critically sampled wavelet transforms, then for undecimated wavelet transforms. Two different optimization criteria are presented: to achieve sparsity by $$L^1$$ -minimization or by $$L^4$$ -norm maximization. Masking can be employed by introducing weighting factors. Examples are given to demonstrate and evaluate the approach.

Aims: Neoadjuvant chemotherapy (NAC) remains an important therapeutic option for advanced oesophageal cancer (OC). Pathological tumour regression grade (TRG) may offer additional information by directing adjuvant treatment and/or follow-up but its clinical value remains unclear. We analysed the prognostic value of TRG and associated pathological factors in OC patients enrolled in the Medical Research Council (MRC) OE02 trial.Methods and results: Histopathology was reviewed in 497 resections from OE02 trial participants randomised to surgery (S group; n = 244) or NAC followed by surgery [chemotherapy plus surgery (CS) group; n = 253]. The association between TRG groups [responders (TRG1-3) versus non-responders (TRG4-5)], pathological lymph node (LN) status and overall survival (OS) was analysed. One hundred and ninety-five of 253 (77%) CS patients were classified as 'non-responders', with a significantly higher mortality risk compared to responders [hazard ratio (HR) = 1.53, 95% confidence interval (CI) = 1.05-2.24, P = 0.026]. OS was significantly better in patients without LN metastases irrespective of TRG [non-responders HR = 1.87, 95% CI = 1.33-2.63, P Conclusion: Lymph node status post-NAC is the most important prognostic factor in patients with resectable oesophageal cancer, irrespective of TRG. Potential clinical implications, e.g. adjuvant treatment or intensified follow-up, reinforce the importance of LN dissection for staging and prognostication.

This paper showcases the setting and results of the first Two-Player General Video Game AI Competition, which ran in 2016 at the IEEE World Congress on Computational Intelligence and the IEEE Conference on Computational Intelligence and Games. The challenges for the general game AI agents are expanded in this track from the single-player version, looking at direct player interaction in both competitive and cooperative environments of various types and degrees of difficulty. The focus is on the agents not only handling multiple problems, but also having to account for another intelligent entity in the game, who is expected to work toward their own goals (winning the game). This other player will possibly interact with first agent in a more engaging way than the environment or any nonplaying character may do. The top competition entries are analyzed in detail and the performance of all agents is compared across the four sets of games. The results validate the competition system in assessing generality, as well as showing Monte Carlo tree search continuing to dominate by winning the overall championship. However, this approach is closely followed by rolling horizon evolutionary algorithms, employed by the winner of the second leg of the contest.

Background It is uncertain whether repeated measurements of a multi-target biomarker panel may help to personalize medical heart failure (HF) therapy to improve outcome in chronic HF. Methods This analysis included 499 patients from the Trial of Intensified versus standard Medical therapy in Elderly patients with Congestive Heart Failure (TIME-CHF), aged ≥ 60 years, LVEF ≤ 45%, and NYHA ≥ II, who had repeated clinical visits within 19 months follow-up. The interaction between repeated measurements of biomarkers and treatment effects of loop diuretics, spironolactone, β-blockers, and renin-angiotensin system (RAS) inhibitors on risk of HF hospitalization or death was investigated in a hypothesis-generating analysis. Generalized estimating equation (GEE) models were used to account for the correlation between recurrences of events in a patient. Results One hundred patients (20%) had just one event (HF hospitalization or death) and 87 (17.4%) had at least two events. Loop diuretic up-titration had a beneficial effect for patients with high interleukin-6 (IL6) or high high-sensitivity C-reactive protein (hsCRP) (interaction, P = 0.013 and P = 0.001), whereas the opposite was the case with low hsCRP (interaction, P = 0.013). Higher dosage of loop diuretics was associated with poor outcome in patients with high blood urea nitrogen (BUN) or prealbumin (interaction, P = 0.006 and P = 0.001), but not in those with low levels of these biomarkers. Spironolactone up-titration was associated with lower risk of HF hospitalization or death in patients with high cystatin C (CysC) (interaction, P = 0.021). β-Blockers up-titration might have a beneficial effect in patients with low soluble fms-like tyrosine kinase-1 (sFlt) (interaction, P = 0.021). No treatment biomarker interactions were found for RAS inhibition. Conclusion The data of this post hoc analysis suggest that decision-making using repeated biomarker measurements may be very promising in bringing treatment of heart failure to a new level in the context of predictive, preventive, and personalized medicine. Clearly, prospective testing is needed before this novel concept can be adopted. Clinical trial registration isrctn.org, identifier: ISRCTN43596477 Electronic supplementary material The online version of this article (10.1007/s13167-018-0137-7) contains supplementary material, which is available to authorized users.

AimTo identify, predict and validate distinct glycaemic trajectories among patients with newly diagnosed type 2 diabetes treated in primary care, as a first step towards more effective patient-centred care.MethodsWe conducted a retrospective study in two cohorts, using routinely collected individual patient data from primary care practices obtained from two large Dutch diabetes patient registries. Participants included adult patients newly diagnosed with type 2 diabetes between January 2006 and December 2014 (development cohort, n=10528; validation cohort, n=3777). Latent growth mixture modelling identified distinct glycaemic 5-year trajectories. Machine learning models were built to predict the trajectories using easily obtainable patient characteristics in daily clinical practice.ResultsThree different glycaemic trajectories were identified: (1) stable, adequate glycaemic control (76.5% of patients); (2) improved glycaemic control (21.3% of patients); and (3) deteriorated glycaemic control (2.2% of patients). Similar trajectories could be discerned in the validation cohort. Body mass index and glycated haemoglobin and triglyceride levels were the most important predictors of trajectory membership. The predictive model, trained on the development cohort, had a receiver-operating characteristic area under the curve of 0.96 in the validation cohort, indicating excellent accuracy.ConclusionsThe developed model can effectively explain heterogeneity in future glycaemic response of patients with type 2 diabetes. It can therefore be used in clinical practice as a quick and easy tool to provide tailored diabetes care.

Metastatic disease remains invariably fatal. Until truly curative therapies are developed, can clinical oncology benefit from lessons learned in pest management?

This paper deals with integration of energy storage systems into electricity markets. We explain why the energy storage systems increase flexibility of both power systems and energy markets and why such flexibility is desirable, particularly when variable renewable energy sources are being used in existing power systems. As opposed to the existing literature, our model includes a dual technology energy storage system, acting in two different markets. We introduce a mathematical formulation for this model applied to two Dutch electricity markets. Adopting optimal control approach with the goal to maximize the yearly benefit, we show that the dual energy storage system can be profitable already when the same buying/selling strategies are adopted for the working days and weekends. We show that the profitability (slightly) increases with different buying/selling strategies for the weekdays and weekends. Finally, we demonstrate how the yearly benefit varies with size and efficiency of the devices chosen and market prices.

The rise of internet and mobile technologies (such as smartphones) provide a harness of data and an opportunity to learn about peoples' states, behavior, and context in regard to several application areas such as health. Eating behavior is an area that can benefit from the development of effective e-coaching applications which utilize psychological theories and data science techniques. In this paper, we propose a framework of how machine learning techniques can effectively be used in order to fully exploit data collected from a mobile application (``Think Slim'') which is designed to assess eating behavior using experience sampling methods. The overall goal is to analyze individual states of a person status (emotions, location, activity, etc.) and assess their impact on unhealthy eating. Building on data collected from different participants, a classification algorithm (decision tree tailored to longitudinal data) is used to warn people prior to a possible unhealthy eating event and a clustering algorithm (hierarchical agglomerative clustering) is used for profiling the participants and generalize for new users of the application. Finally, a framework to offer feedback via adaptive messages (intervention) and recommendations prior to possible unhealthy eating events is presented. Results from applying our methods reveal that participants can be clustered to six robust groups based on their eating behavior and that there are specific rules that discriminate which conditions lead to healthy versus unhealthy eating. Consequently, these rules can be utilized to provide adaptive semi-tailored feedback to users who, through this method, are assisted in learning under which conditions are more prone to unhealthy eating. Effectiveness of the approach is confirmed by observing a decreasing trend in rule activation towards the end of intervention period.

Background Although heart failure (HF) patients are known to experience repeated hospitalizations, most studies evaluated only time to first event. N-Terminal B-type natriuretic peptide (NT-proBNP)–guided therapy has not convincingly been shown to improve HF-specific outcomes, and effects on recurrent all-cause hospitalization are uncertain. Therefore, we investigated the effect of NT-proBNP–guided therapy on recurrent events in HF with the use of a time-between-events approach in a hypothesis-generating analysis. Methods and Results The Trial of Intensified Versus Standard Medical Therapy in Elderly Patients With Congestive Heart Failure (TIME-CHF) randomized 499 HF patients, aged ≥60 years, left ventricular ejection fraction ≤45%, New York Heart Association functional class ≥I,I to NT-proBNP–guided versus symptom-guided therapy for 18 months, with further follow-up for 5.5 years. The effect of NT-proBNP–guided therapy on recurrent HF-related and all-cause hospitalizations and/or all-cause death was explored. One hundred four patients (49 NT-proBNP–guided, 55 symptom-guided) experienced 1 and 275 patients (133 NT-proBNP–guided, 142 symptom-guided) experienced ≥2 all-cause hospitalization events. Regarding HF hospitalization, 132 patients (57 NT-proBNP–guided, 75 symptom-guided) experienced 1 and 122 patients (57 NT-proBNP–guided, 65 symptom-guided) experienced ≥2 events. NT-proBNP–guided therapy was significant in preventing 2nd all-cause hospitalizations (hazard ratio [HR] 0.83; P = .01), in contrast to nonsignificant results in preventing 1st all-cause hospitalization events (HR 0.91; P = .35). This was not the case regarding HF hospitalization events (HR 0.85 [P = .14] vs HR 0.73 [P = .01]) The beneficial effect of NT-proBNP–guided therapy was seen only in patients aged

Objectives The purpose of this study was to evaluate the accuracy of noninvasive reconstructions of epicardial potentials, electrograms, activation and recovery isochrones, and beat origins by simultaneously performing electrocardiographic imaging (ECGI) and invasive epicardial electrography in intact animals. Background Noninvasive imaging of electrical potentials at the epicardium, known as ECGI, is increasingly applied in patients to assess normal and abnormal cardiac electrical activity. Methods Body-surface potentials and epicardial potentials were recorded in normal anesthetized dogs. Computed tomography scanning provided a torso-heart geometry that was used to reconstruct epicardial potentials from body-surface potentials. Results Electrogram reconstructions attained a moderate accuracy compared with epicardial recordings (median correlation coefficient: 0.71), but with considerable variation (interquartile range: 0.36 to 0.86). This variation could be explained by a spatial mismatch (overall resolution was 20 mm) that was most apparent in regions with electrographic transition. More accurate derivation of activation times (Pearson R: 0.82), recovery times (R: 0.73), and the origin of paced beats (median error: 10 mm; interquartile range: 7 to 17 mm) was achieved by a spatiotemporal approach that incorporates the characteristics of the respective electrogram and neighboring electrograms. Reconstruction of beats from repeated single-site pacing showed a stable localization of origin. Cardiac motion, currently ignored in ECGI, correlates negatively with reconstruction accuracy. Conclusions ECGI shows a decent median accuracy, but variability in electrogram reconstruction can be sizable. At present, therefore, clinical interpretations of ECGI should not be made on the basis of single electrograms only. Incorporating local spatiotemporal characteristics allows for accurate reconstruction of epicardial activation and recovery patterns, and beat origin localization to a 10-mm precision. Even more reliable interpretations are expected when the influences of cardiac motion are accounted for in ECGI.

Human activity recognition has received a lot of attention recently, mainly thanks to the advancements in sensing technologies and systems’ increasing computational power. However, complexity in human movements, sensing devices’ noise and person-specific characteristics impose challenges that still remain to be overcome. In the proposed work, a novel, multi-modal human action recognition method is presented for handling the aforementioned issues. Each action is represented by a basis vector and spectral analysis is performed on an affinity matrix of new action feature vectors. Using modality-dependent kernel regressors for computing the affinity matrix, complexity is reduced and robust low-dimensional representations are achieved. The proposed scheme supports online adaptivity of modalities, in a dynamic fashion, according to their automatically inferred reliability. Evaluation on three publicly available datasets demonstrates the potential of the approach.

Many NP-hard problems on graphs become tractable on graphs of low treewidth, but the corresponding algorithms require access to a tree decomposition of the graph of low (ideally, minimum) width. Unfortunately computation of treewidth is itself NP-hard and a wide variety of exact, heuristic and approximation algorithms have been proposed for this problem. To support this ongoing research we present here ToTo, an open graph database for computation, storage and rapid retrieval of tree decompositions. We hope that the database will become both a central repository for important graphs and benchmark datasets and extend the use of treewidth beyond the usual communities: the database and associated algorithms can be accessed via a web browser and do not require installation of any specialist software. (C) 2016 Elsevier B.V. All rights reserved.

Using data generated by extensive simulations of a process where individuals interact globally in a coordination game and iteratively imitate the action of the most successful individual in their local neighbourhood, we study the characteristics of the observation network and distribution of initial choices that facilitate (a speedy) convergence to efficient coordination. Knowing these characteristics is crucial when intervening in social network structures with the intention to nudge society to a socially preferred outcome. The most important factor appears to be the share of individuals that are initially seeded with the socially desirable action. Only for a small window of this share, other factors, including the degree distribution in the network and the segregation of individuals using similar actions, have an influence on the dynamic process of achieving efficient coordination. Moreover, networks possessing the properties of a scale-free network aremore likely to yield an efficient outcome compared to small-world networks.

Research in Artificial Intelligence has shown that machines can be programmed to perform as well as, or even better than humans in specific tasks, such as playing Chess, recognizing images, or driving cars. However, humans still have the advantage of being able to switch between all these tasks autonomously, while a machine needs to be reprogrammed for each new task. The field of Artificial General Intelligence aims at creating a program that is able to perform heterogeneous tasks in different environments autonomously, similarly to humans. The ability to plan and make decisions is one of the core capabilities expected of such a program. My research presents various improvements for a strategy that searches for the best actions to perform in an unknown environment, using board games and video games as test domains. The research shows that the performance of an agent that plays many unknown (video) games can be improved by reasoning on the game rules in a computationally efficient way, using game-independent information to guide the search, and adapting and diversifying the search for each new game being played.

Domains such as logo synthesis, in which the data has a high degree of multi-modality, still pose a challenge for generative adversarial networks (GANs). Recent research shows that progressive training (ProGAN) and mapping network extensions (StyleGAN) enable both increased training stability for higher dimensional problems and better feature separation within the embedded latent space. However, these architectures leave limited control over shaping the output of the network, which is an undesirable trait in the case of logo synthesis. This paper explores a conditional extension to the StyleGAN architecture with the aim of firstly, improving on the low resolution results of previous research and, secondly, increasing the controllability of the output through the use of synthetic class-conditions. Furthermore, methods of extracting such class conditions are explored with a focus on the human interpretability, where the challenge lies in the fact that, by nature, visual logo characteristics are hard to define. The introduced conditional style-based generator architecture is trained on the extracted class-conditions in two experiments and studied relative to the performance of an unconditional model. Results show that, whilst the unconditional model more closely matches the training distribution, high quality conditions enabled the embedding of finer details onto the latent space, leading to more diverse output., accepted for poster presentation at ICMLA 2019, data+code available: https://github.com/cedricoeldorf/ConditionalStyleGAN

We present an algorithm selection framework based on machine learning for the exact computation of treewidth, an intensively studied graph parameter that is NP-hard to compute. Specifically, we analyse the comparative performance of three state-of-the-art exact treewidth algorithms on a wide array of graphs and use this information to predict which of the algorithms, on a graph by graph basis, will compute the treewidth the quickest. Experimental results show that the proposed meta-algorithm outperforms existing methods on benchmark instances on all three performance metrics we use: in a nutshell, it computes treewidth faster than any single algorithm in isolation. We analyse our results to derive insights about graph feature importance and the strengths and weaknesses of the algorithms we used. Our results are further evidence of the advantages to be gained by strategically blending machine learning and combinatorial optimisation approaches within a hybrid algorithmic framework. The machine learning model we use is intentionally simple to emphasise that speedup can already be obtained without having to engage in the full complexities of machine learning engineering. We reflect on how future work could extend this simple but effective, proof-of-concept by deploying more sophisticated machine learning models.

The Digital Ludeme Project (DLP) aims to reconstruct and analyse over 1000 traditional strategy games using modern techniques. One of the key aspects of this project is the development of Ludii, a general game system that will be able to model and play the complete range of games required by this project. Such an undertaking will create a wide range of possibilities for new AI challenges. In this paper we describe many of the features of Ludii that can be used. This includes designing and modifying games using the Ludii game description language, creating agents capable of playing these games, and several advantages the system has over prior general game software., Accepted at the IEEE Conference on Games (CoG) 2019 (Demo paper)

Background: The study of repetitive (quasi-periodic) spatio-temporal patterns in complex dynamical systems with a well defined spatial structure may be complicated if the recurrent behavior is confined to specific local regions, where it lasts for a limited time. This can decrease the efficacy of recurrence plots (RPs) in recognizing such patterns. It then becomes important to first detect whether repetitive spatio-temporal patterns are present, and if so, where they are located (both in space and time), to facilitate a focused RP analysis approach. This study proposes a novel framework for spatio-temporal detection of local recurrence of a quasi-periodic nature in complex dynamical systems. A motivating application for this framework is the analysis of atrial fibrillation to better understand the heart tissue involved.Methods: The spatio-temporal data observed from the system are decomposed by means of principal component analysis to identify the points in the spatial structure exhibiting quasi-periodic recurrent patterns. The frequency content of the principal components is used to determine if such patterns are present, and the corresponding eigenvectors are used to identify the points associated with those components. Geometric information about proximity of these points is used to cluster them into local regions. A sliding temporal window is used to detect the start and end of each pattern.Results: A first simulation shows how the proposed framework can handle multiple recurrent patterns simultaneously occurring in a spatial structure of a dynamical system. A second simulation shows how the method can handle more complicated patterns like 2D nonlinear spiraling waves, typical of many diffusion processes. The framework is then applied to real data to detect recurrent patterns in wave fronts propagating inside the heart during atrial fibrillation. This analysis can unveil regions of recurrence in the atria that were not visible with standard RP analysis.Conclusion: A novel framework for detecting spatio-temporal repetitive patterns in complex dynamical systems is introduced. It allows retrieve the correct recurrences associated with known 2D traveling waves, while the same information is not visible with standard RP analysis. This framework can be effectively used to investigate recurrence in real dynamical systems as cardiac arrhythmia.

In recent years, state-of-the-art game-playing agents often involve policies that are trained in self-playing processes where Monte Carlo tree search (MCTS) algorithms and trained policies iteratively improve each other. The strongest results have been obtained when policies are trained to mimic the search behaviour of MCTS by minimising a cross-entropy loss. Because MCTS, by design, includes an element of exploration, policies trained in this manner are also likely to exhibit a similar extent of exploration. In this paper, we are interested in learning policies for a project with future goals including the extraction of interpretable strategies, rather than state-of-the-art game-playing performance. For these goals, we argue that such an extent of exploration is undesirable, and we propose a novel objective function for training policies that are not exploratory. We derive a policy gradient expression for maximising this objective function, which can be estimated using MCTS value estimates, rather than MCTS visit counts. We empirically evaluate various properties of resulting policies, in a variety of board games., Comment: Accepted at the IEEE Conference on Games (CoG) 2019