Search

In a planar L-drawing of a directed graph (digraph) each edge e is represented as a polyline composed of a vertical segment starting at the tail of e and a horizontal segment ending at the head of e. Distinct edges may overlap, but not cross. Our main focus is on bimodal graphs, i.e., digraphs admitting a planar embedding in which the incoming and outgoing edges around each vertex are contiguous. We show that every plane bimodal graph without 2-cycles admits a planar L-drawing. This includes the class of upward-plane graphs. Finally, outerplanar digraphs admit a planar L-drawing - although they do not always have a bimodal embedding - but not necessarily with an outerplanar embedding., Appears in the Proceedings of the 28th International Symposium on Graph Drawing and Network Visualization (GD 2020)

In this paper we investigate the board game Cram, which isan impartial combinatorial game, using an alpha-beta solver. Since Cram is anotoriously hard game in the sense that it is difficult to obtain reliableand useful domain knowledge to use in the search process, we decided torely on knowledge obtained from Combinatorial Game Theory (CGT).The first and most effective addition to our solver is to incorporateendgame databases prefilled with CGT values (nimbers) for all positionsfitting on boards with at most 30 squares. This together with twoefficient move-ordering heuristics aiming at early splitting positions intofragments fitting in the available databases gives a large improvement ofsolving power. Next we define five more heuristics based on CGT thatcan be used to further reduce the sizes of the search trees considerably.In the final version of our program we were able to solve all odd x oddCram boards for which results were available from the literature (evenx even and odd x even boards are trivially solved). Investigating newboards led to solving two boards not solved before, namely the 3 x 21board, a first-player win, and the 5 x 11 board, a second-player win.

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.

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.

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.

Many of the famous single-player games, commonly called puzzles, can be shown to be NP-Complete. Indeed, this class of complexity contains hundreds of puzzles, since people particularly appreciate completing an intractable puzzle, such as Sudoku, but also enjoy the ability to check their solution easily once it's done. For this reason, using constraint programming is naturally suited to solve them. In this paper, we focus on logic puzzles described in the Ludii general game system and we propose using the XCSP formalism in order to solve them with any CSP solver., Comment: Accepted at the IEEE Conference on Games (CoG) 2019 (Short paper)

This paper proposes a regime-switching recurrent network model (RS-RNN) for non-stationary time series. The RS-RNN model emits a mixture density with dynamic nonlinear regimes that fit flexibly data distributions with non-Gaussian shapes. The key novelties are: development of an original representation of the means of the component distributions by dynamic nonlinear recurrent networks, and derivation of a corresponding expectation maximization (EM) training algorithm for finding the model parameters. The elaborated switching dynamic nonlinear regimes make the RS-RNN especially attractive for describing non-stationary environmental time series. The results show that the RS-RNN applied to a real-world wind speed time series achieves standardized residuals similar to popular previous models, but it is more accurate distribution forecasting than other linear switching (MS-AR) and nonlinear neural network (MLP and RNN) models.

In the network design game with n players, every player chooses a path in an edge-weighted graph to connect her pair of terminals, sharing costs of the edges on her path with all other players fairly. It has been shown that the price of stability of any network design game is at most $$H_n$$ , the n-th harmonic number. This bound is tight for directed graphs. For undirected graphs, it has only recently been shown that the price of stability is at most $$H_n \left( 1-\frac{1}{\Theta (n^4)} \right) $$ , while the worst-case known example has price of stability around 2.25. We improve the upper bound considerably by showing that the price of stability is at most $$H_{n/2} + \varepsilon $$ for any $$\varepsilon $$ starting from some suitable $$n \ge n(\varepsilon )$$ . We also study quality measures of different solution concepts for the multicast network design game on a ring topology. We recall from the literature a lower bound of $$\frac{4}{3}$$ and prove a matching upper bound for the price of stability. Therefore, we answer an open question posed by Fanelli et al. (Theor Comput Sci 562:90–100, 2015). We prove an upper bound of 2 for the ratio of the costs of a potential optimizer and of an optimum, provide a construction of a lower bound, and give a computer-assisted argument that it reaches 2 for any precision. We then turn our attention to players arriving one by one and playing myopically their best response. We provide matching lower and upper bounds of 2 for the myopic sequential price of anarchy (achieved for a worst-case order of the arrival of the players). We then initiate the study of myopic sequential price of stability and for the multicast game on the ring we construct a lower bound of $$\frac{4}{3}$$ , and provide an upper bound of $$\frac{26}{19}$$ . To the end, we conjecture and argue that the right answer is $$\frac{4}{3}$$ .

We study the task of gathering k energy-constrained mobile agents in an undirected edge-weighted graph. Each agent is initially placed on an arbitrary node and has a limited amount of energy, which constrains the distance it can move. Since this may render gathering at a single point impossible, we study three variants of near-gathering:the goal is to move the agents into a configuration that minimizes either (i) the radius of a ball containing all agents, (ii) the maximum distance between any two agents, or (iii) the average distance between the agents. We prove that (i) is polynomial-time solvable, (ii) has a polynomial-time 2-approximation with a matching np-hardness lower bound, while (iii) admits a polynomial-time \(2(1-\tfrac{1}{k})\)-approximation, but no fptas, unless \(\text {p}=\text {np}\). We extend some of our results to additive approximation.keywordsmobile agentspower-aware robotslimited batterygatheringgraph algorithmsapproximationcomputational complexity.

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

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.

Ludii is a new general game system, currently under development, which aims to support a wider range of games than existing systems and approaches. It is being developed primarily for the task of game design, but offers a number of other potential benefits for game and AI researchers, professionals and hobbyists. This paper is based on an interactive demonstration of Ludii at thuis year’s Advances in Computer Games conference (ACG 2019). It describes the approach behind Ludii, how it works, how it is used, and what it can potentially do.

Inspired by the recent social movement of #MeToo, we are building a chatbot to assist survivors of sexual harassment cases (designed for the city of Maastricht but can easily be extended). The motivation behind this work is twofold: properly assist survivors of such events by directing them to appropriate institutions that can offer them help and increase the incident documentation so as to gather more data about harassment cases which are currently under reported. We break down the problem into three data science/machine learning components: harassment type identification (treated as a classification problem), spatio-temporal information extraction (treated as Named Entity Recognition problem) and dialogue with the users (treated as a slot-filling based chatbot). We are able to achieve a success rate of more than 98% for the identification of a harassment-or-not case and around 80% for the specific type harassment identification. Locations and dates are identified with more than 90% accuracy and time occurrences prove more challenging with almost 80%. Finally, initial validation of the chatbot shows great potential for the further development and deployment of such a beneficial for the whole society tool.

Currently, in all augmented reality (AR) or virtual reality (VR) educational experiences, the evolution of the experience (game, exercise or other) and the assessment of the user's performance are based on her/his (re)actions which are continuously traced/sensed. In this paper, we propose the exploitation of the sensors available in the AR/VR systems to enhance the current AR/VR experiences, taking into account the users' affect state that changes in real time. Adapting the difficulty level of the experience to the users' affect state fosters their engagement which is a crucial issue in educational environments and prevents boredom and anxiety. The users' cues are processed enabling dynamic user profiling. The detection of the affect state based on different sensing inputs, since diverse sensing devices exist in different AR/VR systems, is investigated, and techniques that have been undergone validation using state-of-the-art sensors are presented.

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.

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.

A novel solution to the problem whether to transfer based on exchangeablity test.Statistically testing if the source data is generated from the target distribution.The test is non-parametric and distribution free.Empirically justified the proposed test is effective for predicting transfer result. This paper introduces a non-parametric test to decide whether to transfer data from a source domain to a target domain to improve the generalization performance of predictive models on the target domain. The test is based on the conformal prediction framework: it statistically tests whether the target and source data are generated from the same distribution under the exchangeability assumption. The experiments show that the test is capable of outperforming existing methods when it decides on instance transfer.

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.

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.

Ludii is a general game system being developed as part of the ERC-funded Digital Ludeme Project (DLP). While its primary aim is to model, play, and analyse the full range of traditional strategy games, Ludii also has the potential to support a wide range of AI research topics and competitions. This paper describes some of the future competitions and challenges that we intend to run using the Ludii system, highlighting some of its most important aspects that can potentially lead to many algorithm improvements and new avenues of research. We compare and contrast our proposed competition motivations, goals and frameworks against those of existing general game playing competitions, addressing the strengths and weaknesses of each platform., Accepted at the IEEE Conference on Games (CoG) 2019

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

We consider the problem of collectively delivering a package from a specified source to a designated target location in a graph, using multiple mobile agents. Each agent starts from a distinct vertex of the graph, and can move along the edges of the graph carrying the package. However, each agent has limited energy budget allowing it to traverse a path of bounded length b; thus, multiple agents need to collaborate to move the package to its destination. Given the positions of the agents in the graph and their energy budgets, the problem of finding a feasible movement schedule is called the collaborative delivery problem and has been studied before.one of the open questions from previous results is what happens when the delivery must follow a fixed path given in advance. Although this special constraint reduces the search space for feasible solutions, the problem of finding a feasible schedule remains np hard (as the original problem). We consider the optimization version of the problem that asks for the optimal energy budget b per agent which allows for a feasible delivery schedule, given the initial positions of the agents. We show the existence of better approximations for the fixed-path version of the problem (at least for the restricted case of single pickup per agent), compared to the known results for the general version of the problem, thus answering the open question from the previous paper.we provide polynomial time approximation algorithms for both directed and undirected graphs, and establish hardness of approximation for the directed case. Note that the fixed path version of collaborative delivery requires completely different techniques since a single agent may be used multiple times, unlike the general version of collaborative delivery studied before. We show that restricting each agent to a single pickup allows better approximations for fixed path collaborative delivery compared to the original problem. Finally, we provide a polynomial time algorithm for determining a feasible delivery strategy, if any exists, for a given budget b when the number of available agents is bounded by a constant.

Although General Game Playing (GGP) systems can facilitate useful research in Artificial Intelligence (AI) for game-playing, they are often computationally inefficient and somewhat specialised to a specific class of games. However, since the start of this year, two General Game Systems have emerged that provide efficient alternatives to the academic state of the art -- the Game Description Language (GDL). In order of publication, these are the Regular Boardgames language (RBG), and the Ludii system. This paper offers an experimental evaluation of Ludii. Here, we focus mainly on a comparison between the two new systems in terms of two key properties for any GGP system: simplicity/clarity (e.g. human-readability), and efficiency., Accepted at the IEEE Conference on Games (CoG) 2019 (Short paper)

Virtual assistants, also known as intelligent conversational systems such as Google's Virtual Assistant and Apple's Siri, interact with human-like responses to users' queries and finish specific tasks. Meanwhile, existing recommendation technologies model users' evolving, diverse and multi-aspect preferences to generate recommendations in various domains/applications, aiming to improve the citizens' daily life by making suggestions. The repertoire of actions is no longer limited to the one-shot presentation of recommendation lists, which can be insufficient when the goal is to offer decision support for the user, by quickly adapting to his/her preferences through conversations. Such an interactive mechanism is currently missing from recommendation systems. This article sheds light on the gap between virtual assistants and recommendation systems in terms of different technological aspects. In particular, we try to answer the most fundamental research question, which are the missing technological factors to implement a personalized intelligent conversational agent for producing accurate recommendations while taking into account how users behave under different conditions. The goal is, instead of adapting humans to machines, to actually provide users with better recommendation services so that machines will be adapted to humans in daily life.

This paper proposes using a linear function approximator, rather than a deep neural network (DNN), to bias a Monte Carlo tree search (MCTS) player for general games. This is unlikely to match the potential raw playing strength of DNNs, but has advantages in terms of generality, interpretability and resources (time and hardware) required for training. Features describing local patterns are used as inputs. The features are formulated in such a way that they are easily interpretable and applicable to a wide range of general games, and might encode simple local strategies. We gradually create new features during the same self-play training process used to learn feature weights. We evaluate the playing strength of an MCTS player biased by learnt features against a standard upper confidence bounds for trees (UCT) player in multiple different board games, and demonstrate significantly improved playing strength in the majority of them after a small number of self-play training games., Comment: Accepted at IEEE CEC 2019, Special Session on Games. Copyright of final version held by IEEE

Venue recommendation strategies are built upon collaborative filtering techniques that rely on matrix factorisation (mf), to model users’ preferences. Various cross-domain strategies have been proposed to enhance the effectiveness of mf-based models on a target domain, by transferring knowledge from a source domain. Such cross-domain recommendation strategies often require user overlap, that is common users on the different domains. However, in practice, common users across different domains may not be available. To tackle this problem, recently, several cross-domains strategies without users’ overlaps have been introduced. In this paper, we investigate the performance of state-of-the-art cross-domain recommendation that do not require overlap of users for the venue recommendation task on three large location-based social networks (lbsn) datasets. Moreover, in the context of cross-domain recommendation we extend a state-of-the-art sequential-based deep learning model to boost the recommendation accuracy. Our experimental results demonstrate that state-of-the-art cross-domain recommendation does not clearly contribute to the improvements of venue recommendation systems, and, further we validate this result on the latest sequential deep learning-based venue recommendation approach. Finally, for reproduction purposes we make our implementations publicly available.keywordscross-domain recommendationvenue suggestiontransfer learning.

Nowadays, users open multiple accounts on social media platforms and e-commerce sites, expressing their personal preferences on different domains. However, users' behaviors change across domains, depending on the content that users interact with, such as movies, music, clothing and retail products. The main challenge is how to capture users' complex preferences when generating cross-domain recommendations, that is exploiting users' preferences from source domains to generate recommendations in a target domain. In this study, we propose a Neural Attentive Cross-domain model, namely NAC. We design a neural architecture, to carefully transfer the knowledge of user preferences across domains by taking into account the cross-domain latent effects of multiple source domains on users' selections in a target domain. In addition, we introduce a cross-domain behavioral attention mechanism to adaptively perform the weighting of users' preferences from the source domains, and consequently generate accurate cross-domain recommendations. Our experiments on ten cross-domain recommendation tasks show that the proposed NAC model achieves higher recommendation accuracy than other state-of-the-art methods for both ordinary and cold-start users. Furthermore, we study the effect of the proposed cross-domain behavioral attention mechanism and show that it is a key factor to our model's performance.

Recently, significant improvements have been achieved in various natural language processing tasks using neural sequence-to-sequence models. While aiming for the best generation quality is important, ultimately it is also necessary to develop models that can assess the quality of their output. In this work, we propose to use the similarity between training and test conditions as a measure for models' confidence. We investigate methods solely using the similarity as well as methods combining it with the posterior probability. While traditionally only target tokens are annotated with confidence measures, we also investigate methods to annotate source tokens with confidence. By learning an internal alignment model, we can significantly improve confidence projection over using state-of-the-art external alignment tools. We evaluate the proposed methods on downstream confidence estimation for machine translation (MT). We show improvements on segment-level confidence estimation as well as on confidence estimation for source tokens. In addition, we show that the same methods can also be applied to other tasks using sequence-to-sequence models. On the automatic speech recognition (ASR) task, we are able to find 60% of the errors by looking at 20% of the data., Comment: 8 pages; INLG 2019

Predicting salient regions in natural images requires the detection of objects that are present in a scene. To develop robust representations for this challenging task, high-level visual features at multiple spatial scales must be extracted and augmented with contextual information. However, existing models aimed at explaining human fixation maps do not incorporate such a mechanism explicitly. Here we propose an approach based on a convolutional neural network pre-trained on a large-scale image classification task. The architecture forms an encoder-decoder structure and includes a module with multiple convolutional layers at different dilation rates to capture multi-scale features in parallel. Moreover, we combine the resulting representations with global scene information for accurately predicting visual saliency. Our model achieves competitive and consistent results across multiple evaluation metrics on two public saliency benchmarks and we demonstrate the effectiveness of the suggested approach on five datasets and selected examples. Compared to state of the art approaches, the network is based on a lightweight image classification backbone and hence presents a suitable choice for applications with limited computational resources, such as (virtual) robotic systems, to estimate human fixations across complex natural scenes., Comment: Accepted Manuscript

Exploiting the selections of social friends and foes can efficiently face the data scarcity of user preferences and the cold-start problem. In this paper, we present a Social Deep Pairwise Learning model, namely SDPL. According to the Bayesian Pairwise Ranking criterion, we design a loss function with multiple ranking criteria based on the selections of users, and those in their friends and foes to improve the accuracy in the top-k recommendation task. We capture the nonlinearity in user preferences and the social information of trust and distrust relationships by designing a deep learning architecture. In each backpropagation step, we perform social negative sampling to meet the multiple ranking criteria of our loss function. Our experiments on a benchmark dataset from Epinions, among the largest publicly available that has been reported in the relevant literature, demonstrate the effectiveness of the proposed approach, outperforming other state-of-the art methods. In addition, we show that our deep learning strategy plays an important role in capturing the nonlinear associations between user preferences and the social information of trust and distrust relationships, and demonstrate that our social negative sampling strategy is a key factor in SDPL.CCS CONCEPTS • Information systems → Collaborative and social computing systems and tools.

In this study we investigate the recommendation problem with multifaceted contextual information to overcome the scarcity of users' check-in data in Location-based Social Networks. To generate accurate personalized Point-of-Interest (POI) recommendations in the presence of data scarcity, we account for both users' and POIs' contextual information such as the social influence of friends, as well as the geographical and sequential transition influence of POIs on user's check-in behavior. We first propose a multi-view learning strategy to capture the multifaceted contextual information of users and POIs along with users' check-in data. Then, we feed the learned user and POI latent vectors to a deep neural framework, to capture their non-linear correlations. Finally, we formulate the objective function of our geo-based deep collaborative filtering model (GeoDCF) as a Bayesian personalized ranking problem to focus on the top-k recommendation task and we learn the parameters of our model via backpropagation. Our experiments on real-world datasets confirm that GeoDCF achieves high recommendation accuracy, significantly outperforming other state-of-the-art methods. Furthermore, we confirm the influence of both users' and POIs' contextual information on our GeoDCF model. The evaluation datasets are publicly available at: http://snap.stanford.edu/data/loc-gowalla.html, https://sites.google.com/site/yangdingqi/home/foursquare-dataset.

Recent studies have shown that incorporating users' reviews into the collaborative filtering strategy can significantly boost the recommendation accuracy. A pressing challenge resides on learning how reviews influence users' rating behaviors. In this paper, we propose an Adversarial Training approach for Review-based recommendations, namely ATR. We design a neural architecture of sequence-to-sequence learning to calculate the deep representations of users' reviews on items following an adversarial training strategy. At the same time we jointly learn to factorize the rating matrix, by regularizing the deep representations of reviews with the user and item latent features. In doing so, our model captures the non-linear associations among reviews and ratings while producing a review for each user-item pair. Our experiments on publicly available datasets demonstrate the effectiveness of the proposed model, outperforming other state-of-the-art methods.

ARM has become the leading processor architecture for mobile and IoT devices, while it has recently started claiming a bigger slice of the server market pie as well. As such, it will not be long before malware more regularly target the ARM architecture. Therefore, the stealthy operation of Virtual Machine Introspection (VMI) is an obligation to successfully analyze and proactively mitigate this growing threat. Stealthy VMI has proven itself perfectly suitable for malware analysis on Intel's architecture, yet, it often lacks the foundation required to be equally effective on ARM.In this paper, we closely examine both ARMv7 and ARMv8 architectures to identify shortcomings and develop novel techniques necessary for effective virtualization-based dynamic malware analysis. We implement and open-source a prototype, named altp2m, for the open source Xen Project hypervisor on ARM. Compared to traditional VMI approaches, our solution enables hypervisors to dynamically allocate and switch among multiple guest memory views by utilizing the Second Level Address Translation (SLAT). Further, we implement an alternative single-stepping mechanism and leverage the execute-only capability of the SLAT mechanism on ARMv8 to enable stealthy in-guest instrumentation. To target also ARMv7-based systems, we manipulate the TLB organization through altp2m to coordinate the guest kernel execution flow. To demonstrate the effectiveness of our system, we combine all building blocks of our work to form the foundation for the dynamic malware analysis system DRAKVUF on ARM. Overall, our experiments reveal that our novel dynamic analysis system is stealthy, efficient, and is perfectly suited to assist malware analysts to quickly comprehend the behavior and reduce the mitigation time of malware targeting ARM.

A difficult aspect of a time dependent classification task is that the data are not IID sampled. To model this dependency several approaches in longitudinal analysis were developed. However, these approaches either have trouble estimating their generalization performance or are parametric in a statistical sense. To overcome these problems we propose in this paper a new approach of time dependent ensembles. Our approach decomposes the time dependent classification task into a series of classification tasks with IID sampled data. Each task can be solved by a classifier that is not supposed to model any dependency in the data. This allows for the use of a much broader spectrum of existing approaches than is possible on the original data. The classifiers associated with the tasks form the time dependent ensembles. The ensembles estimates the final class of the objects being classified by using a voting scheme. The experiments show the potential of the time dependent ensembles.

A new approach for unknown event detection in videos with dense motion, such as crowds or dynamic textures, is developed, without requiring the estimation of optical flow, with no prior knowledge about normal or abnormal events, and with no training. The proposed method directly extracts motion statistics from the phase of the video's Fourier transform and detects changes in them, and in the video, by applying sequential statistical change detection theory. Focus is placed on the motion component, as videos of densely moving entities, such as temporal textures and crowds, often have a very similar appearance, but different dynamic features. Experiments with synthetically generated datasets demonstrate the method's operation under various conditions, while experiments on a recently introduced crowd dataset show that it succeeds at detecting new events in videos of crowds, with no training, and no prior knowledge of the location of new events in space and time.

The speed of game rules processing plays an essential role in the performance of a General Game Playing (GGP) agent. Propositional Networks (propnets) are an example of a highly efficient representation of game rules. So far, in GGP, only software implementations of propnets have been proposed and investigated. In this paper, we present the first implementation of propnets on Field-Programmable Gate Arrays (FPGAs), showing that they perform between 25 and 58 times faster than a software-propnet for most of the tested games. We also integrate the FPGA-propnet within an MCTS agent, discussing the challenges of the process, and possible solutions for the identified shortcomings.

The cross-conformal prediction is an approach to confidence region prediction. It provides a trade-off between the validity and informational efficiency of the prediction regions from one hand and the computational complexity from another. In this paper we introduce a new cross-conformal approach based on ensembles. The new approach is more computationally efficient and provides gains in the validity and informational efficiency of the prediction regions. Hence, it is a good candidate for big data (analytics) when prediction regions with confidence values are required.

In this paper we present a combinatorial game-theoretic analysis of special Domineering positions. In particular we investigate complex positions that are aggregates of simpler fragments, linked via bridging squares. We build on two theorems that exploit the characteristic of an aggregate of two fragments having as game-theoretic value the sum of the values of the fragments. We extend these theorems to deal with the case of multiple-connected networks with an arbitrary number of fragments, possibly including cycles. As an application, we introduce several interesting Domineering positions, dubbed Snowflake molecules. A Snowflake molecule consists of a core subfragment with some "hooks" attached. We investigate four different cores, all with a different type of value: 1) the CrystalBall core, with number value 0; 2) the BigI core, with number value 1; 3) the DoubleT core, with nimber value *2; and 4) the GearWheel core, with switch value +/- 1*. We then show how from each such Snowflake molecule chains of Snowflakes can be built (a kind of polymerization) with known values, including flat networks of Snowflakes (crystallization).

General Game Playing (GGP) aims at creating computer programs able to play any arbitrary game at an expert level given only its rules. The lack of game-specific knowledge and the necessity of learning a strategy online have made Monte-Carlo Tree Search (MCTS) a suitable method to tackle the challenges of GGP. An efficient search-control mechanism can substantially increase the performance of MCTS. The RAVE strategy and its more recent variant, GRAVE, have been proposed for this reason. In this paper we further investigate the use of GRAVE for GGP and compare its performance with the more established RAVE strategy and with a new variant, called HRAVE, that uses more global information. Experiments show that for some games GRAVE and HRAVE perform better than RAVE, with GRAVE being the most promising one overall.

Simultaneous move games model discrete, multistage interactions where at each stage players simultaneously choose their actions. At each stage, a player does not know what action the other player will take, but otherwise knows the full state of the game. This formalism has been used to express games in general game playing and can also model many discrete approximations of real-world scenarios. In this paper, we describe both novel and existing algorithms that compute strategies for the class of two-player zero-sum simultaneous move games. The algorithms include exact backward induction methods with efficient pruning, as well as Monte Carlo sampling algorithms. We evaluate the algorithms in two different settings: the offline case, where computational resources are abundant and closely approximating the optimal strategy is a priority, and the online search case, where computational resources are limited and acting quickly is necessary. We perform a thorough experimental evaluation on six substantially different games for both settings. For the exact algorithms, the results show that our pruning techniques for backward induction dramatically improve the computation time required by the previous exact algorithms. For the sampling algorithms, the results provide unique insights into their performance and identify favorable settings and domains for different sampling algorithms.

Affective computing researchers adopt a variety of methods in analysing or synthesizing aspects of human behaviour. The choice of method depends on which behavioural cues are considered salient or straightforward to capture and comprehend, as well as the overall context of the interaction. Thus, each approach focuses on modelling certain information and results to dedicated representations. However, analysis or synthesis is usually done by following label-based representations, which usually have a direct mapping to a feature vector. The goal of the presented work is to introduce an interim representational mechanism that associates low-level gesture expressivity parameters with a high-level dimensional representation of affect. More specifically, it introduces a novel methodology for associating easily extracted, low-level gesture data to the affective dimensions of activation and evaluation. For this purpose, a user perception test was carried out in order to properly annotate a dataset, by asking participants to assess each gesture in terms of the perceived activation (active/passive) and evaluation (positive/negative) levels. In affective behaviour modelling, the contribution of the proposed association methodology is twofold: On one hand, when analysing affective behaviour, it can enable the fusion of expressivity parameters alongside with any other modalities coded in higher-level affective representations, leading, in this way, to scalable multimodal analysis. On the other hand, it can enforce the process of synthesizing composite human behaviour (e.g. facial expression, gestures and body posture) since it allows for the translation of dimensional values of affect into synthesized expressive gestures.

Given a set of directed paths (called lines) l, a public transportation network is a directed graph g_l=(v_l,a_l)g_l=(v_l,a_l) which contains exactly the vertices and arcs of every line l\in ll\in l. An st-route is a pair (\pi ,\gamma )(\pi ,\gamma ) where \gamma =\langle l_1,\ldots ,l_h \rangle \gamma =\langle l_1,\ldots ,l_h \rangle is a line sequence and \pi \pi is an st-path in g_lg_l which is the concatenation of subpaths of the lines l_1,\ldots ,l_hl_1,\ldots ,l_h, in this order. Given a threshold \beta \beta , we present an algorithm for listing all st-paths \pi \pi for which a route (\pi ,\gamma )(\pi ,\gamma ) with |\gamma | \le \beta |\gamma | \le \beta exists, and we show that the running time of this algorithm is polynomial with respect to the input and the output size. We also present an algorithm for listing all line sequences \gamma \gamma with |\gamma |\le \beta |\gamma |\le \beta for which a route (\pi ,\gamma )(\pi ,\gamma ) exists, and show how to speed it up using preprocessing. Moreover, we show that for the problem of finding an st-route (\pi ,\gamma )(\pi ,\gamma ) that minimizes the number of different lines in \gamma \gamma , even computing an o(\log |v|)o(\log |v|)-approximation is np-hard.

Designing a logo is a long, complicated, and expensive process for any designer. However, recent advancements in generative algorithms provide models that could offer a possible solution. Logos are multi-modal, have very few categorical properties, and do not have a continuous latent space. Yet, conditional generative adversarial networks can be used to generate logos that could help designers in their creative process. We propose LoGAN: an improved auxiliary classifier Wasserstein generative adversarial neural network (with gradient penalty) that is able to generate logos conditioned on twelve different colors. In 768 generated instances (12 classes and 64 logos per class), when looking at the most prominent color, the conditional generation part of the model has an overall precision and recall of 0.8 and 0.7 respectively. LoGAN's results offer a first glance at how artificial intelligence can be used to assist designers in their creative process and open promising future directions, such as including more descriptive labels which will provide a more exhaustive and easy-to-use system., 6 page, ICMLA18

In neural machine translation (NMT), it has become standard to translate using subword units to allow for an open vocabulary and improve accuracy on infrequent words. Byte-pair encoding (BPE) and its variants are the predominant approach to generating these subwords, as they are unsupervised, resource-free, and empirically effective. However, the granularity of these subword units is a hyperparameter to be tuned for each language and task, using methods such as grid search. Tuning may be done inexhaustively or skipped entirely due to resource constraints, leading to sub-optimal performance. In this paper, we propose a method to automatically tune this parameter using only one training pass. We incrementally introduce new BPE vocabulary online based on the held-out validation loss, beginning with smaller, general subwords and adding larger, more specific units over the course of training. Our method matches the results found with grid search, optimizing segmentation granularity while significantly reducing overall training time. We also show benefits in training efficiency and performance improvements for rare words due to the way embeddings for larger units are incrementally constructed by combining those from smaller units.

Urbanism is no longer planned on paper thanks to powerful models and 3D simulation platforms. However, current work is not open to the public and lacks an optimisation agent that could help in decision making. This paper describes the creation of an open-source simulation based on an existing Dutch liveability score with a built-in AI module. Features are selected using feature engineering and Random Forests. Then, a modified scoring function is built based on the former liveability classes. The score is predicted using Random Forest for regression and achieved a recall of 0.83 with 10-fold cross-validation. Afterwards, Exploratory Factor Analysis is applied to select the actions present in the model. The resulting indicators are divided into 5 groups, and 12 actions are generated. The performance of four optimisation algorithms is compared, namely NSGA-II, PAES, SPEA2 and backslashepsilon -MOEA, on three established criteria of quality: cardinality, the spread of the solutions, spacing, and the resulting score and number of turns. Although all four algorithms show different strengths, backslashepsilon -MOEA is selected to be the most suitable for this problem. Ultimately, the simulation incorporates the model and the selected AI module in a GUI written in the Kivy framework for Python. Tests performed on users show positive responses and encourage further initiatives towards joining technology and public applications.

This paper introduces novel deep architectures using the hybrid neural-kernel core model as the first building block. The proposed models follow a combination of a neural networks based architecture and a kernel based model enriched with pooling layers. In particular, in this context three kernel blocks with average, maxout and convolutional pooling layers are introduced and examined. We start with a simple merging layer which averages the output of the previous representation layers. The maxout layer on the other hand triggers competition among different representations of the input. Thanks to this pooling layer, not only the dimensionality of the output of multi-scale representations is reduced but also multiple sub-networks are formed within the same model. In the same context, the pointwise convolutional layer is also employed with the aim of projecting the multi-scale representations onto a new space. Experimental results show an improvement over the core deep hybrid model as well as kernel based models on several real-life datasets. (C) 2019 Elsevier Ltd. All rights reserved.