In non-linear systems, where explicit analytic solutions usually can't be found, visualisation is a powerful approach which can give insights into the dynamical behaviour of models; it is also crucial for teaching this area of mathematics. In this paper we present new software, Fireflies, which exploits the power of graphical processing unit (GPU) computing to produce spectacular interactive visualisations of arbitrary systems of ordinary differential equations. In contrast to typical phase portraits, Fireflies draws the current position of trajectories (projected onto 2D or 3D space) as single points of light, which move as the system is simulated. Due to the massively parallel nature of GPU hardware, Fireflies is able to simulate millions of trajectories in parallel (even on standard desktop computer hardware), producing "swarms" of particles that move around the screen in real-time according to the equations of the system. Particles that move forwards in time reveal stable attractors (e.g. fixed points and limit cycles), while the option of integrating another group of trajectories backwards in time can reveal unstable objects (repellers). Fireflies allows the user to change the parameters of the system as it is running, in order to see the effect that they have on the dynamics and to observe bifurcations. We demonstrate the capabilities of the software with three examples: a two-dimensional "mean field" model of neuronal activity, the classical Lorenz system, and a 15-dimensional model of three interacting biologically realistic neurons.

Insider threats are one of the most damaging risk factors for the IT systems and infrastructure of a company or an organization; identification of insider threats has prompted the interest of the world academic research community, with several solutions having been proposed to alleviate their potential impact. For the implementation of the experimental stage described in this study, the Convolutional Neural Network (from now on CNN) algorithm was used and implemented via the Google TensorFlow program, which was trained to identify potential threats from images produced by the available dataset. From the examination of the images that were produced and with the help of Machine Learning, the question of whether the activity of each user is classified as malicious or not for the Information System was answered.

We introduce form factors that relate the amplitude of a QED process in vacuum to its corresponding background-field process. The latter is characterised by a reduced S-matrix element where one or more photon field operators are replaced by classical background fields. In the associated Feynman diagram, external photon lines are supplanted with lines representing the c-number field. This modifies the cross section by factors proportional to powers of the Fourier amplitude of the classical field (and its complex conjugate). We demonstrate this explicitly by comparing different reaction channels of low-energy photon-photon scattering in a classical background. We find that background field cross sections typically undergo coherent enhancement and for some reaction channels display a more favourable scaling with centre-of-mass energy compared to the vacuum process. Similar coherent enhancement may be found for leading-order pair annihilation to one photon, but this competes with kinematic suppression. This suppression can be minimised by using an x-ray free electron laser as the classical background.

Phase transitions in a non-perturbative regime can be studied by ab initio Lattice Field Theory methods. The status and future research directions for LFT investigations of Quantum Chromo-Dynamics under extreme conditions are reviewed, including properties of hadrons and of the hypothesized QCD axion as inferred from QCD topology in different phases. We discuss phase transitions in strong interactions in an extended parameter space, and the possibility of model building for Dark Matter and Electro-Weak Symmetry Breaking. Methodological challenges are addressed as well, including new developments in Artificial Intelligence geared towards the identification of different phases and transitions.

Group emotion recognition in the wild is a challenging problem, due to the unstructured environments in which everyday life pictures are taken. Some of the obstacles for an effective classification are occlusions, variable lighting conditions, and image quality. In this work we present a solution based on a novel combination of deep neural networks and Bayesian classifiers. The neural network works on a bottom-up approach, analyzing emotions expressed by isolated faces. The Bayesian classifier estimates a global emotion integrating top-down features obtained through a scene descriptor. In order to validate the system we tested the framework on the dataset released for the Emotion Recognition in the Wild Challenge 2017. Our method achieved an accuracy of 64.68% on the test set, significantly outperforming the 53.62% competition baseline.

Landing an unmanned aerial vehicle (UAV) on a ground marker is an open problem despite the effort of the research community. Previous attempts mostly focused on the analysis of hand-crafted geometric features and the use of external sensors in order to allow the vehicle to approach the land-pad. In this article, we propose a method based on deep reinforcement learning that only requires low-resolution images taken from a down-looking camera in order to identify the position of the marker and land the UAV on it. The proposed approach is based on a hierarchy of Deep Q-Networks (DQNs) used as high-level control policy for the navigation toward the marker. We implemented different technical solutions, such as the combination of vanilla and double DQNs, and a partitioned buffer replay. Using domain randomization we trained the vehicle on uniform textures and we tested it on a large variety of simulated and real-world environments. The overall performance is comparable with a state-of-the-art algorithm and human pilots.

In Wang-Landau type algorithms, Monte-Carlo updates are performed with respect to the density of states, which is iteratively refined during simulations. The partition function and thermodynamic observables are then obtained by standard integration. In this work, our recently introduced method in this class (the LLR approach) is analysed and further developed. Our approach is a histogram free method particularly suited for systems with continuous degrees of freedom giving rise to a continuum density of states, as it is commonly found in Lattice Gauge Theories and in some Statistical Mechanics systems. We show that the method possesses an exponential error suppression that allows us to estimate the density of states over several orders of magnitude with nearly-constant {\it relative} precision. We explain how ergodicity issues can be avoided and how expectation values of arbitrary observables can be obtained within this framework. We then demonstrate the method using Compact U(1) Lattice Gauge Theory. A thorough study of the algorithm parameter dependence of the results is performed and compared with the analytically expected behaviour. We obtain high precision values for the critical coupling for the phase transition and for the peak value of the specific heat for lattice sizes ranging from $8^4$ to $20^4$. Our results perfectly agree with the reference values reported in the literature, which covers lattice sizes up to $18^4$. Robust results for the $20^4$ volume are obtained for the first time. This latter investigation, which, due to strong metastabilities developed at the pseudo-critical coupling, so far has been out of reach even on supercomputers with importance sampling approaches, has been performed to high accuracy with modest computational resources. Other situations where the method is expected to be superior to importance sampling techniques are pointed out.

Whether monochromatic, pulsed, or even constant and crossed, the field used to describe the interaction of charged fermions with an intense laser beam is mainly assumed to be of plane-wave form. We consider a simple extension to plane-wave fields and consider a scalar particle in a non-lightlike, univariate and transverse propagating electromagnetic wave. The existence of some known exact solutions in this case allows us to analyse various proposed approximations in the literature as well as the plane wave model. The results also describe some of the quantum dynamics of a scalar particle in a standing wave background.

All lattice-QCD calculations of the hadronic-vacuum-polarization contribution to the muon's anomalous magnetic moment to-date have been performed with degenerate up- and down-quark masses. Here we calculate directly the strong-isospin-breaking correction to $a_\mu^{\rm HVP}$ for the first time with physical values of $m_u$ and $m_d$ and dynamical $u$, $d$, $s$, and $c$ quarks, thereby removing this important source of systematic uncertainty. We obtain a relative shift to be applied to lattice-QCD results obtained with degenerate light-quark masses of $\delta a_\mu^{{\rm HVP,} m_u \neq m_d}$= +1.5(7)%, in agreement with estimates from phenomenology and a recent lattice-QCD calculation with unphysically heavy pions.

The problem of gas detection is relevant to many real-world applications, such as leak detection in industrial settings and landfill monitoring. Using mobile robots for gas detection has several advantages and can reduce danger for humans. In our work, we address the problem of planning a path for a mobile robotic platform equipped with a remote gas sensor, which minimizes the time to detect all gas sources in a given environment. We cast this problem as a coverage planning problem by defining a basic sensing operation -- a scan with the remote gas sensor -- as the field of "view" of the sensor. Given the computing effort required by previously proposed offline approaches, in this paper we suggest a online coverage algorithm, called Next-Best-Smell, adapted from the Next-Best-View class of exploration algorithms. Our algorithm evaluates candidate locations with a global utility function, which combines utility values for travel distance, information gain, and sensing time, using Multi-Criteria Decision Making. In our experiments, conducted both in simulation and with a real robot, we found the performance of the Next-Best-Smell approach to be comparable with that of the state-of-the-art offline algorithm, at much lower computational cost.

The density-of-states method (Phys.Rev.Lett. 109 (2012) 111601) features an exponential error suppression and is not restricted to theories with positive probabilistic weight. It is applied to the SU(2) gauge theory at finite densities of heavy quarks. The key ingredient here is the Polyakov line probability distribution, which is obtained of over 80 orders of magnitude. We briefly address whether the exponential error suppression could be sufficient to simulate theories with a strong sign problem.

The continued evolution and diversity of malware constitutes a major threat in modern systems. It is well proven that security defenses currently available are ineffective to mitigate the skills and imagination of cyber-criminals necessitating the development of novel solutions. Deep learning algorithms and artificial intelligence (AI) are rapidly evolving with remarkable results in many application areas. Following the advances of AI and recognizing the need for efficient malware detection methods, this paper presents a new approach for malware detection based on binary visualization and self-organizing incremental neural networks. The proposed method's performance in detecting malicious payloads in various file types was investigated and the experimental results showed that a detection accuracy of 91.7% and 94.1% was achieved for ransomware in .pdf and .doc files respectively. With respect to other formats of malicious code and other file types, including binaries, the proposed method behaved well with an incremental detection rate that allows efficiently detecting unknown malware at real-time.

Human Action Recognition is an important task of Human Robot Interaction as cooperation between robots and humans requires that artificial agents recognise complex cues from the environment. A promising approach is using trained classifiers to recognise human actions through sequences of skeleton poses extracted from images or RGB-D data from a sensor. However, with many different data-sets focused on slightly different sets of actions and different algorithms it is not clear which strategy produces highest accuracy for indoor activities performed in a home environment. This work discussed, tested and compared classic algorithms, namely, support vector machines and k-nearest neighbours, to 2 similar hierarchical neural gas approaches, the growing when required neural gas and the growing neural gas.

We provide in this paper a tutorial and a comprehensive survey of QoE management solutions in current and future networks. We start with a high level description of QoE management for multimedia services, which integrates QoE modelling, monitoring, and optimization. This followed by a discussion of HTTP Adaptive Streaming (HAS) solutions as the dominant technique for streaming videos over the best-effort Internet. We then summarize the key elements in SDN/NFV along with an overview of ongoing research projects, standardization activities and use cases related to SDN, NFV, and other emerging applications. We provide a survey of the state-of-the-art of QoE management techniques categorized into three different groups: a) QoE-aware/driven strategies using SDN and/or NFV; b) QoE-aware/driven approaches for adaptive streaming over emerging architectures such as multi-access edge computing, cloud/fog computing, and information-centric networking; and c) extended QoE management approaches in new domains such as immersive augmented and virtual reality, mulsemedia and video gaming applications. Based on the review, we present a list of identified future QoE management challenges regarding emerging multimedia applications, network management and orchestration, network slicing and collaborative service management in softwarized networks. Finally, we provide a discussion on future research directions with a focus on emerging research areas in QoE management, such as QoE-oriented business models, QoE-based big data strategies, and scalability issues in QoE optimization.

We explore some consequences of the crossing symmetry for defect conformal field theories, focusing on codimension one defects like flat boundaries or interfaces. We study surface transitions of the 3d Ising and other O(N) models through numerical solutions to the crossing equations with the method of determinants. In the extraordinary transition, where the low-lying spectrum of the surface operators is known, we use the bootstrap equations to obtain information on the bulk spectrum of the theory. In the ordinary transition the knowledge of the low-lying bulk spectrum allows to calculate the scale dimension of the relevant surface operator, which compares well with known results of two-loop calculations in 3d. Estimates of various OPE coefficients are also obtained. We also analyze in 4-epsilon dimensions the renormalization group interface between the O(N) model and the free theory and check numerically the results in 3d.

We prove that the existence spectrum of Mendelsohn triple systems whose associated quasigroups satisfy distributivity corresponds to the Loeschian numbers, and provide some enumeration results. We do this by considering a description of the quasigroups in terms of commutative Moufang loops. In addition we provide constructions of Mendelsohn quasigroups that fail distributivity for as many combinations of elements as possible. These systems are analogues of Hall triple systems and anti-mitre Steiner triple systems respectively.

Human physical motion activity identification has many potential applications in various fields, such as medical diagnosis, military sensing, sports analysis, and human-computer security interaction. With the recent advances in smartphones and wearable technologies, it has become common for such devices to have embedded motion sensors that are able to sense even small body movements. This study collected human activity data from 60 participants across two different days for a total of six activities recorded by gyroscope and accelerometer sensors in a modern smartphone. The paper investigates to what extent different activities can be identified by utilising machine learning algorithms using approaches such as majority algorithmic voting. More analyses are also provided that reveal which time and frequency domain based features were best able to identify individuals motion activity types. Overall, the proposed approach achieved a classification accuracy of 98 percent in identifying four different activities: walking, walking upstairs, walking downstairs, and sitting while the subject is calm and doing a typical desk-based activity.

Finite density quantum field theories have evaded first principle Monte-Carlo simulations due to the notorious sign-problem. The partition function of such theories appears as the Fourier transform of the generalised density-of-states, which is the probability distribution of the imaginary part of the action. With the advent of Wang-Landau type simulation techniques and recent advances, the density-of-states can be calculated over many hundreds of orders of magnitude. Current research addresses the question whether the achieved precision is high enough to reliably extract the finite density partition function, which is exponentially suppressed with the volume. In my talk, I review the state-of-play for the high precision calculations of the density-of-states as well as the recent progress for obtaining reliable results from highly oscillating integrals. I will review recent progress for the $Z_3$ quantum field theory for which results can be obtained from the simulation of the dual theory, which appears to free of a sign problem.

In this paper we use counting arguments to prove that the expected percentage coverage of a $d$ dimensional parameter space of size $n$ when performing $k$ trials with either Latin Hypercube sampling or Orthogonal sampling (when $n=p^d$) is the same. We then extend these results to an experimental design setting by projecting onto a 2 dimensional subspace. In this case the coverage is equivalent to the Orthogonal sampling setting when the dimension of the parameter space is two. These results are confirmed by simulations. The ideas presented here have particular relevance when attempting to perform uncertainty quantification or when building populations of models.