Search

QC 20240917

QC 20240917

AI-assisted food intake monitoring systems have drawn considerable attention from researchers. To date, various approaches have been proposed to objectively and unobtrusively detect food intake activities by utilizing novel sensors and machine learning techniques. In the development of automated food intake monitoring systems, one crucial step is to evaluate the generated results from machine learning models. In this study, we illustrate the challenge arising from the inefficiency of traditional sliding-window-based evaluation in translating results into clinical indices (i.e. number of bites). Additionally, existing evaluation metrics only focus on detection performance (count the occurrence of eating gestures); however, the segmentation performance (temporal boundary of eating gesture) is missed, which is also a clinically meaningful index. Apart from the discussion of existing evaluation methods in food intake monitoring, we introduce the segment-wise evaluation scheme using the Intersection Over Union (IoU) as threshold to assess performance. This method facilitates the evaluation of both the detection and segmentation performance of eating activities. Two public food intake datasets are used in our case study to illustrate that the segment-wise method can yield more detailed information and a more comprehensive evaluation when compared to existing metrics. The proposed evaluation scheme has the potential to be applied to other human activity recognition (HAR) cases.

Reinforcement Learning aims at identifying and evaluating efficient control policies from data. In many real-world applications, the learner is not allowed to experiment and cannot gather data in an online manner (this is the case when experimenting is expensive, risky or unethical). For such applications, the reward of a given policy (the target policy) must be estimated using historical data gathered under a different policy (the behavior policy). Most methods for this learning task, referred to as Off-Policy Evaluation (OPE), do not come with accuracy and certainty guarantees. We present a novel OPE method based on Conformal Prediction that outputs an interval containing the true reward of the target policy with a prescribed level of certainty. The main challenge in OPE stems from the distribution shift due to the discrepancies between the target and the behavior policies. We propose and empirically evaluate different ways to deal with this shift. Some of these methods yield conformalized intervals with reduced length compared to existing approaches, while maintaining the same certainty level., QC 20240326

In evolutionary algorithms and metaheuristics, defining when applying a specific operator is important. Besides, in complex optimization problems, multiple populations can be used to explore the search space simultaneously. However, one of the main problems is extracting information from the populations and using it to evolve the solutions. This article presents the inequality-based multi-population differential evo-lution (IMDE). This algorithm uses the K-means to generate subpopulations (settlements). Two variables are extracted from the settlements, the diversity and the Gini index, which measure the solutions' distribution and the solutions' inequality regarding fitness. The Gini index and the diversity are used in the IMDE to dynamically modify the scalation factor and the crossover rate. Experiments over a set of benchmark functions with different degrees of complexity validate the performance of the IMDE. Besides comparisons, statistical and ranking average validate the search capabilities of the IMDE.

In this paper, we consider a privacy signaling game problem for binary alphabets and single-bit transmission where a transmitter has a pair of messages, one of which is a casual message that needs to be conveyed, whereas the other message contains sensitive data and needs to be protected. The receiver wishes to estimate both messages to acquire as much information as possible. For this setup, we study the interactions between the transmitter and the receiver with non-aligned information-theoretic objectives (modeled by mutual information and hamming distance) due to the privacy concerns of the transmitter. We derive conditions under which Nash and/or Stackelberg equilibria exist and identify the optimal responses of the encoder and decoders strategies for each type of game. One particularly surprising result is that when both types of equilibria exist, they admit the same encoding and decoding strategies. We corroborate our analysis with simulation studies., Part of proceedings: ISBN 978-3-907144-07-7, QC 20221101

Web pages aggressively track users for a variety of purposes from targeted advertisements to enhanced authentication. As browsers move to restrict traditional cookie-based tracking, web pages increasingly move to tracking based on browser fingerprinting. Unfortunately, the state-of-the-art to detect fingerprinting in browsers is often error-prone, resorting to imprecise heuristics and crowd-sourced filter lists. This paper presents EssentialFP, a principled approach to detecting fingerprinting on the web. We argue that the pattern of (i) gathering information from a wide browser API surface (multiple browser-specific sources) and (ii) communicating the information to the network (network sink) captures the essence of fingerprinting. This pattern enables us to clearly distinguish fingerprinting from similar types of scripts like analytics and polyfills. We demonstrate that information flow tracking is an excellent fit for exposing this pattern. To implement EssentialFP we leverage, extend, and deploy JSFlow, a state-of-the-art information flow tracker for JavaScript, in a browser. We illustrate the effectiveness of EssentialFP to spot fingerprinting on the web by evaluating it on two categories of web pages: one where the web pages perform analytics, use polyfills, and show ads, and one where the web pages perform authentication, bot detection, and fingerprinting-enhanced Alexa top pages.

Relevant research has shown the potential to understand the game user experience (GUX) more accurately and reliably by measuring the user’s psychophysiological responses. However, the current studies are still very scarce and limited in scope and depth. Besides, the low-detection accuracy and the common use of the professional physiological signal apparatus make it difficult to be applied in practice. This article analyzes the GUX, particularly flow experience, based on users’ physiological responses, including the galvanic skin response (GSR) and heart rate (HR) signals, captured by low-cost wearable devices. Based on the collected data sets regarding two test games and the mixed data set, several classification models were constructed to detect the flow state automatically. Hereinto, two strategies were proposed and applied to improve classification performance. The results demonstrated that the flow experience of game users could be effectively classified from other experiences. The best accuracies of two-way classification and three-way classification under the support of the proposed strategies were over 90% and 80%, respectively. Specifically, the comparison test with the existing results showed that Strategy1 could significantly reduce the negative interference of individual differences in physiological signals and improve the classification accuracy. In addition, the results of the mixed data set identified the potential of a general classification model of flow experience., © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.ISSN: CD: 2372-2541Funding Agency:10.13039/501100001809-National Natural Science Foundation of China; Scientific and Technological Innovation Foundation of Shunde Graduate School, USTB

One way to analyse the behaviour of machine learning models is through local explanations that highlight input features that maximally influence model predictions. Sensitivity analysis, which involves analysing the effect of input perturbations on model predictions, is one of the methods to generate local explanations. Meaningful input perturbations are essential for generating reliable explanations, but there exists limited work on what such perturbations are and how to perform them. This work investigates these questions in the context of machine listening models that analyse audio. Specifically, we use a state-of-the-art deep singing voice detection (SVD) model to analyse whether explanations from SoundLIME (a local explanation method) are sensitive to how the method perturbs model inputs. The results demonstrate that SoundLIME explanations are sensitive to the content in the occluded input regions. We further propose and demonstrate a novel method for quantitatively identifying suitable content type(s) for reliably occluding inputs of machine listening models. The results for the SVD model suggest that the average magnitude of input mel-spectrogram bins is the most suitable content type for temporal explanations., QC 20210420

Vector Symbolic Architectures belong to a family of related cognitive modeling approaches that encode symbols and structures in high-dimensional vectors. Similar to human subjects, whose capacity to process and store information or concepts in short-term memory is subject to numerical restrictions, the capacity of information that can be encoded in such vector representations is limited and one way of modeling the numerical restrictions to cognition. In this paper, we analyze these limits regarding information capacity of distributed representations. We focus our analysis on simple superposition and more complex, structured representations involving convolutive powers to encode spatial information. In two experiments, we find upper bounds for the number of concepts that can effectively be stored in a single vector only depending on the dimensionality of the underlying vector space., QC 20210420

Unsupervised learning of hierarchical representations has been one of the most vibrant research directions in deep learning during recent years. In this work we study biologically inspired unsupervised strategies in neural networks based on local Hebbian learning. We propose new mechanisms to extend the Bayesian Confidence Propagating Neural Network (BCPNN) architecture, and demonstrate their capability for unsupervised learning of salient hidden representations when tested on the MNIST dataset., QC 20210419

Data-driven functions for operation and management often require measurements collected through monitoring for model training and prediction. The number of data sources can be very large, which requires a significant communication and computing overhead to continuously extract and collect this data, as well as to train and update the machine-learning models. We present an online algorithm, called OSFS, that selects a small feature set from a large number of available data sources, which allows for rapid, low-overhead, and effective learning and prediction. OSFS is instantiated with a feature ranking algorithm and applies the concept of a stable feature set, which we introduce in the paper. We perform extensive, experimental evaluation of our method on data from an in-house testbed. We find that OSFS requires several hundreds measurements to reduce the number of data sources by two orders of magnitude, from which models are trained with acceptable prediction accuracy. While our method is heuristic and can be improved in many ways, the results clearly suggests that many learning tasks do not require a lengthy monitoring phase and expensive offline training., QC 20210303

We present a method to automatically find security strategies for the use case of intrusion prevention. Following this method, we model the interaction between an attacker and a defender as a Markov game and let attack and defense strategies evolve through reinforcement learning and self-play without human intervention. Using a simple infrastructure configuration, we demonstrate that effective security strategies can emerge from self-play. This shows that self-play, which has been applied in other domains with great success, can be effective in the context of network security. Inspection of the converged policies show that the emerged policies reflect common-sense knowledge and are similar to strategies of humans. Moreover, we address known challenges of reinforcement learning in this domain and present an approach that uses function approximation, an opponent pool, and an autoregressive policy representation. Through evaluations we show that our method is superior to two baseline methods but that policy convergence in self-play remains a challenge., QC 20210302

SNOW 3G is one of the core algorithms for confidentiality and integrity in several 3GPP wireless communication standards, including the new Next Generation (NG) 5G. It is believed to be resistant to classical cryptanalysis. In this paper, we show that SNOW 3G can be broken by a fault attack based on bitstream modification. By changing the content of some look-up tables in the bitstream, we reduce the non-linear state updating function of SNOW 3G to a linear one. As a result, it becomes possible to recover the key from a known plaintext-ciphertext pair. To our best knowledge, this is the first successful bitstream modification attack on SNOW 3G., QC 20210302

The extensive use of Simulation-based Serious Games (SSGs) has made a revolution in educational techniques. As a potentially significant feature for SSG design and evaluation, the term fidelity (the similarity between an SSG and its real reference) emerges and attracts increasing attention. The study of fidelity not only benefits the design, development, and analysis of an SSG with the consideration of improving the learning effect but also contributes to the investment reduction of an SSG. However, the term fidelity is used inconsistently in current literature. The introduction of new technologies (e.g. virtual reality) and the blend of multiform SSGs also facilitate the extension of fidelity with new connotations. All lead to confusing concepts and vague measure metrics. Besides, the relationship between fidelity and learning effect is still uncertain. A new vision and a comprehensive conceptual framework of fidelity for more general applications are in need. In this paper, further exploration and discussion of these issues in relation to fidelity of SSGs are presented through a systematic review. A general conceptual framework considering both aspects of the SSG system itself and the learners is developed and applied to analyze fidelity in SSGs. Based on that, a discussion on fidelity related issues of SSG design and development is presented.

For low dimensional classification problems we propose the novel DIOPT approach which considers the construction of a discretized feature space. Predictions for all cells in this space are obtained by means of a reference classifier and the class labels are stored in a lookup table generated by enumerating the complete space. This then leads to extremely high classification throughput as inference consists only of discretizing the relevant features and reading the class label from the lookup table index corresponding to the concatenation of the discretized feature bin indices. Since the size of the lookup table is limited due to memory constraints, the selection of optimal features and their respective discretization levels is paramount. We propose a particular supervised discretization approach striving to achieve maximal class separation of the discretized features, and further employ a purpose-built memetic algorithm to search towards the optimal selection of features and discretization levels. The inference run time and classification accuracy of DIOPT is compared to benchmark random forest and decision tree classifiers in several publicly available data sets. Orders of magnitude improvements are recorded in classification runtime with insignificant or modest degradation in classification accuracy for many of the evaluated binary classification tasks.

In this paper, we propose a robust solution for the path planning and scheduling of the moving objects in a Track-based Traffic Control System (TTCS). The moving objects in a TTCS pass over pre-specified sub-tracks. Each sub-track accommodates at most one moving object in-transit. Due to the uncertainties in the context of a TTCS, we assign an arrival time window to each moving object for each sub-track in its route, instead of an exact value. The moving object can safely enter into the sub-track in the mentioned time window. To develop a safe plan, we adapt the tagged-signal model and provide a rigorous mathematical formalism for the actor model of a TTCS. To illustrate the applicability of the provided semantics, we provide a formal model of TTCSs in the Alloy language and use its analyzer to verify the developed model against system safety properties.

In networked systems engineering, operational data gathered from sensors or logs can be used to build data-driven functions for performance prediction, anomaly detection, and other operational tasks. The number of data sources used for this purpose determines the dimensionality of the feature space for learning and can reach millions for medium-sized systems. Learning on a space with high dimensionality generally incurs high communication and computational costs for the learning process. In this work, we apply and compare a range of methods, including, feature selection, Principle Component Analysis (PCA), and autoencoders with the objective to reduce the dimensionality of the feature space while maintaining the prediction accuracy when compared with learning on the full space. We conduct the study using traces gathered from a test-bed at KTH that runs a video-on-demand service and a key-value store under dynamic load. Our results suggest the feasibility of reducing the dimensionality of the feature space of operational data significantly, by one to two orders of magnitude in our scenarios, while maintaining prediction accuracy. The findings confirm the Manifold Hypothesis in machine learning, which states that real-world data sets tend to occupy a small subspace of the full feature space. In addition, we investigate the tradeoff between prediction accuracy and prediction overhead, which is crucial for applying the results to operational systems., QC 20201111

Software repositories such as Git have become a relevant source of information for software engineer researchers. For instance, the detection of commits that fulfill a given criterion (e.g., bugfixing commits) is one of the most frequent tasks done to understand the software evolution. However, to our knowledge, there is no open-source tool that, given a Git repository, returns all the instances of a given code change pattern. In this paper we present Coming, a tool that takes as input a Git repository and mines instances of code change patterns present on each commit. For that, Coming computes fine-grained code changes between two consecutive revisions, analyzes those changes to determine if they correspond to an instance of a change pattern (specified by the user using XML), and finally, after analyzing all the commits, it presents a) the frequency of code changes and b) the instances found in each commit. We evaluate Coming on a set of 28 pairs of revisions from Defects4J, finding instances of change patterns that involve If conditions on 26 of them., QC 20200402

Identifying relation from clinical texts is a complex and challenging task due to the specific biomedical knowledge. Existing methods for this work generally have the misclassification problem caused by sample class imbalance. In this paper, we propose a hierarchical attention-based convolutional long short-term memory (ConvLSTM) network model to solve this problem. We construct a sentence as multi-dimensional hierarchical sequence and directly learn local and global context information by a single-layer ConvLSTM network. Besides, a hierarchical attention-based pooling is built to capture the parts of a sentence that are relevant with the target semantic relation. Experiments on the 2010 i2b2/VA relation dataset show that our model outperforms several previous state-of-the-art models without relying on any external features., QC 20200622

We study a two-stage identification problem with pre-processing to enable efficient data retrieval and reconstruc- tion. In the enrollment phase, users’ data are stored into the database in two layers. In the identification phase an observer obtains an observation, which originates from an unknown user in the enrolled database through a memoryless channel. The observation is sent for processing in two stages. In the first stage, the observation is pre-processed, and the result is then used in combination with the stored first layer information in the database to output a list of compatible users to the second stage. Then the second step uses the information of users contained in the list from both layers and the original observation sequence to return the exact user identity and a corresponding reconstruction sequence. The rate-distortion regions are characterized for both discrete and Gaussian scenarios. Specifically, for a fixed list size and distortion level, the compression-identification trade-off in the Gaussian scenario results in three different operating cases characterized by three auxiliary functions. While the choice of the auxiliary random variable for the first layer information is essentially unchanged when the identification rate is varied, the second one is selected based on the dominant function within those three. Due to the presence of a mixture of discrete and continuous random variables, the proof for the Gaussian case is highly non-trivial, which makes a careful measure theoretic analysis necessary. In addition, we study a connection of the previous setting to a two observer identification and a related problem with a lower bound for the list size, where the latter is motivated from privacy concerns., QC 20191114. QC 20200318

In this paper we revisit the connections of the distributed hypothesis testing against independence (HT) problem with the Wyner-Ahlswede-Korner (WAK) problem and the identification systems (ID). We show that the strong converse for the WAK problem is equivalent to the strong converse for the HT problem via constructive and nonconstructive transformations of codes. As another consequence of the transformation we provide a new exponentially strong converse equivalence statement. Applying the same idea, we prove a new result that the epsilon-identification capacity of the ID problem is equal to the maximum epsilon-exponent of type II of error in the HT problem when both side compression is allowed., QC 20191114. QC 20200318

The identification of novel biological factors associated with thrombin generation, a key biomarker of the coagulation process, remains a relevant strategy to disentangle pathophysiological mechanisms underlying the risk of venous thrombosis (VT). As part of the MARseille THrombosis Association Study (MARTHA), we measured whole blood DNA methylation levels, plasma levels of 300 proteins, 3 thrombin generation biomarkers (endogeneous thrombin potential, peak and lagtime), clinical and genetic data in 700 patients with VT. The application of a novel high-dimensional multi-levels statistical methodology we recently developed, the data driven sparse Partial Least Square method (ddsPLS), on the MARTHA datasets enabled us 1/ to confirm the role of a known mutation of the variability of endogenous thrombin potential and peak, 2/ to identify a new signature of 7 proteins strongly associated with lagtime., QC 20200121

We investigate the use of entropy-regularized optimal transport (EOT) cost in developing generative models to learn implicit distributions. Two generative models are proposed. One uses EOT cost directly in an one-shot optimization problem and the other uses EOT cost iteratively in an adversarial game. The proposed generative models show improved performance over contemporary models on scores of sample based test., QC 20191001

The capacity of symmetric private information retrieval (PIR) with N servers and K messages, each coded by an (N, M)-MDS code has been characterized as CMDS-SPIR = 1- M/N. A critical assumption for this result is that the randomness is similarly coded by an (N, M)-MDS code, i.e., the code parameters of the messages and randomness are matched. In this work, we are interested in the mismatched case, and as a preliminary result, we establish the capacity of the mismatched MDS coded symmetric PIR (SPIR) problem under an extreme setting, where the messages are coded by an (N, M)-MDS code and the randomness is replicated (i.e., coded by an (N, 1)-MDS code). The capacity is shown to be Cmis-MDS-SPIR = (1 - 1/N). (1 + M-1/N (1+ M/N + . . . (M/N)(K-2)))(-1). Interestingly, Cmis-MDS-SPIR > CMDS-SPIR, so mismatched coded randomness (with more redundancy) is strictly beneficial. Further, mismatched SPIR exhibits properties that are similar to PIR., QC 20191114

We investigate the behavior of the mutual information between two Boolean functions of correlated binary strings. The covariance of these functions is found to be a crucial parameter in the aforementioned mutual information. We then apply this result in the analysis of a specific privacy problem where a user observes a random binary string. Under particular conditions, we characterize the optimal strategy for communicating the outcomes of a function of said string while preventing to leak any information about a different function., QC 20191114

The internet of things (IoT) system integrates heterogeneous systems using centric services to provide a single open solution to process sensor data. During the whole life cycle of IoT systems, developers face the problems of interface management and data interoperability led by increasing complexity. Such problems decrease the efficiency of IoT system development and implementation, such as interface configurations for domain specific systems are difficult if there is not a unified specification. This paper proposed an Open Services for Lifecycle Collaboration (OSLC) approach supporting IoT system development and implementation. The approach integrates domain specific data across the whole lifecycle including development models and sensor data. Moreover, it enables interface management for IoT system development and real-time monitoring for implementations. From the case study, we find an OSLC-based tool, Datalinks, supports data integration and interface management which improves the development efficiency and data interoperability of IoT systems. The integrated data based on OSLC acts the mid-wares of data exchange between physical space and virtual space of IoT system. Moreover, the OSLC-based interfaces are developed based on unified specifications whose reusability is promoted for the future development., QC 20190930

ElectroEncephaloGram (EEG) spectral power has been extensively used to classify Mental Imagery (MI) of movements involving different body parts. However, there is an increasing need to enable classification of MI of movements within the same limb. In this work, EEG spectral power was recorded in seven subjects while they performed MI of closing (grip) and opening (extension of fingers) the hand. The EEG data was analyzed and the feasibility of classifying MI of the two movements were investigated using two different classification algorithms, a linear regression and a Convolutional Neural Network (CNN). Results show that only the CNN is able to significantly classify MI of opening and closing of the hand with an average classification accuracy of 60.4%. This indicates the presence of higher-order non-linear discriminatory information and demonstrates the potential of using CNN in classifying MI of same-limb movements.

The Census Transform (CT) is a well proven method for stereo vision that provides robust matching, with respect to object boundaries, outliers and radiometric distortion, at a low computational cost. Recent CT methods propose patterns for pixel comparison and sparsity, to increase matching accuracy and reduce resource requirements. However, these methods are bounded with respect to symmetry and/or edge length. In this paper, a Genetic algorithm (GA) is applied to find a new and powerful CT method. The proposed method, Genetic Algorithm Census Transform (GACT), is compared with the established CT methods, showing better results for benchmarking datasets. Additional experiments have been performed to study the search space and the correlation between training and evaluation data.

Many problems in operations research require that constraints be specified in the model. Determining right constraints is a hard and laborsome task. We propose an approach to automate this process using artificial intelligence and machine learning principles. We focus on personnel rostering and scheduling problems in which there are often past schedules available and show that it is possible to automatically learn constraints from such examples. To realize this, we adapted some techniques from the constraint programming community and extended them in order to cope with multidimensional examples. The method uses a tensor representation of the example, which helps in capturing the dimensionality as well as the structure of the example, and applies tensor operations to find the constraints that are satisfied by the example. The algorithm also identifies inherent clusters in the data and uses it as background knowledge to learn more detailed constraints. To evaluate the proposed algorithm, we used constraints from the Nurse Rostering Competition and generated solutions that satisfy these constraints; these solutions were then used as examples to learn constraints. Experiments demonstrate that the proposed algorithm is capable of producing human readable constraints that capture the underlying characteristics of the examples., Funding Agency:European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme 694980

For spatial modulation (SM) systems that utilize multiple transmit antennas/patterns with a single radio front-end, we propose a learning approach to predict the average symbol error rate (SER) conditioned on the instantaneous channel state. We show that the predicted SER can he used to lower the average SER over Rayleigh fading channels by selecting the optimal codebook in each transmission instance. Further by exploiting that feedforward artificial neural networks (ANNs) trained with a mean squared error (MSE) criterion estimate the conditional a posteriori probabilities, we maximize the expected rate for each transmission instance and thereby improve the link spectral efficiency., QC 20190603

The problem of time-constrained multi-agent task scheduling and control synthesis is addressed. We assume the existence of a high level plan which consists of a sequence of cooperative tasks, each of which is associated with a deadline and several Quality-of-Service levels. By taking into account the reward and cost of satisfying each task, a novel scheduling problem is formulated and a path synthesis algorithm is proposed. Based on the obtained plan, a distributed hybrid control law is further designed for each agent. Under the condition that only a subset of the agents are aware of the high level plan, it is shown that the proposed controller guarantees the satisfaction of time constraints for each task. A simulation example is given to verify the theoretical results., QC 20190305

In this paper we address the problem of self-triggered control of nonlinear systems under actuator delays. In particular, for globally asymptotically stabilizable systems we exploit the Lipschitz properties of the system's dynamics, and present a self-triggered strategy that guarantees the stability of the sampled closed-loop system with bounded actuator delays., QC 20190305

We investigate flexible service chain mapping in server-centric optical terconnects, handling virtual network function (VNF) dependency operly. Blocking probability decreases by a factor of 10 when signing multiple VNFs in the same server is allowed., QC 20190326

Musical onset detection is a key component in any beat tracking system. Existing onset detection methods are based on temporal/spectral analysis, or methods that integrate temporal and spectral information together with statistical estimation and machine learning models. In this paper, we propose a method to localize onset components in music by using the S-transform, and thus, the method is purely based on temporal/spectral data. Unlike the other methods based on temporal/spectral data, which usually rely on the short time Fourier transform (STET), our method enables effective isolation of crucial frequency subbands due to the frequency dependent resolution of S-transform. Moreover, numerical results show, even with less computationally intensive steps, the proposed method can closely resemble the performance of more resource intensive statistical estimation based approaches., QC 20190603

We consider the control design problem of optimizing the H-2 performance of a closed-loop system despite the presence of a malicious covert attacker. It is assumed that the attacker has incomplete knowledge on the true process we are controlling. To account for this uncertainty, we employ different measures of risk from the so called family of coherent measures of risk. In particular, we compare the closed-loop performance when a nominal value is used, with three different measures of risk: average risk, worst-case scenario and conditional valueat- risk (CVaR). Additionally, applying the approach from a previous work, we derive a convex formulation for the control design problem when CVaR is employed to quantify the risk. A numerical example illustrates the advantages of our approach., QC 20190305, CERCES

We study a hypothesis testing problem to decide whether or not an observation sequence is related to one of users in a database which contains compressed versions of users' data. Our main interest lies on the characterization of the exponent of the probability of the second kind of error when the number of users in the database grows exponentially. We show a lower bound on the error exponent and identify special cases where the bound is tight. Next, we study the c-achievable error exponent and show a sub-region where the lower bound is tight., QC 20190610. QC 20200318

Playing recorded speech samples of an enrolled speaker – “replay attack” – is a simple approach to bypass an automatic speaker ver- ification (ASV) system. The vulnerability of ASV systems to such attacks has been acknowledged and studied, but there has been no research into what spoofing detection systems are actually learning to discriminate. In this paper, we analyse the local behaviour of a replay spoofing detection system based on convolutional neural net- works (CNNs) adapted from a state-of-the-art CNN (LC N NF F T ) submitted at the ASVspoof 2017 challenge. We generate tempo- ral and spectral explanations for predictions of the model using the SLIME algorithm. Our findings suggest that in most instances of spoofing the model is using information in the first 400 milliseconds of each audio instance to make the class prediction. Knowledge of the characteristics that spoofing detection systems are exploiting can help build less vulnerable ASV systems, other spoofing detection systems, as well as better evaluation databases., QC 20190423

Internet of Things provides an ability to interact with, share the data, and expand the capabilities of the physical world in terms of computation, communication, and key control with humans through many new modalities devices in the connected network. Though the availability of the information and performance are higher at lower cost, the usage of such system becomes more complex with the advancement of technologies. The traditional ways like lecture-based and role-playing learning has developed onesided learning and also expensive for the low-income people to acquire such knowledge. On the other hand, serious gaming has helped the users in acquiring new experiences and complex knowledge which are acquired through solving presented challenges whereby the user applies competency to solve these problems. This paper proposes serious gaming as a learning environment for gaining competence, knowledge, and experiences in IoT and knowledge sharing for the users. Moreover, the design of a serious game, effectiveness of ATMSG framework and evaluation results are also discussed., QC 20190305

For linear time-varying systems with possibly constrained states, we propose a hybrid observer that guarantees the containment of the estimated state variables in a prescribed domain of interest. The hybrid observer employs a Kalmantype continuous estimator during the flows while, during the jumps, projects the state estimates onto the set described by the constraint equation. A suitable choice of the flow and jump sets allows to conclude uniform global asymptotic stability of the zero estimation error set., QC 20190305

With the aggressive scaling of submicron technology, intermittent faults are becoming one of the limiting factors in achieving a high reliability in Network-on-Chip (NoC). Increasing test frequency is necessary to detect intermittent faults, which in turn interrupts the execution of applications. On the other hand, the main goal of traditional mapping algorithms is to allocate applications to the NoC platform, ignoring about the test requirement. In this paper, we propose a novel testing-aware mapping algorithm (TAMA) for NoC, targeting intermittent faults on the paths between crossbars. In this approach, the idle links are identified and the components between two crossbars are tested when the application is mapped to the platform. The components can be tested if there is enough time from when the application leaves the platform and a new application enters it. The mapping algorithm is tuned to give a higher priority to the tested paths in the next application mapping. This leaves enough time to test the links and the belonging components that have not been tested in the expected time. Experiment results show that the proposed testing-aware mapping algorithm leads to a significant improvement over FF, NN, CoNA, and WeNA., QC 20180404

The steady decline of IP transit prices in the past two decades has helped fuel the growth of traffic demands in the Internet ecosystem. Despite the declining unit pricing, bandwidth costs remain significant due to ever-increasing scale and reach of the Internet, combined with the price disparity between the Internet's core hubs versus remote regions. In the meantime, cloud providers have been auctioning underutilized computing resources in their marketplace as spot instances for a much lower price, compared to their on-demand instances. This state of affairs has led the networking community to devote extensive efforts to cloud-assisted networks - the idea of offloading network functionality to cloud platforms, ultimately leading to more flexible and highly composable network service chains. We initiate a critical discussion on the economic and technological aspects of leveraging cloud-assisted networks for Internet-scale interconnections and data transfers. Namely, we investigate the prospect of constructing a large-scale virtualized network provider that does not own any fixed or dedicated resources and runs atop several spot instances. We construct a cloud-assisted overlay as a virtual network provider, by leveraging third-party cloud spot instances. We identify three use case scenarios where such approach will not only be economically and technologically viable but also provide performance benefits compared to current commercial offerings of connectivity and transit providers., QC 20191202

Electric power distribution systems encounter fluctuations in supply due to renewable sources with high variability in generation capacity. It is therefore necessary to provide algorithms that are capable of dynamically finding approximate solutions. We propose two semi-distributed algorithms based on ADMM and discuss their advantages and disadvantages. One of the algorithms computes a feasible approximate of the optimal power allocation at each time instance. We require coordination between the nodes to guarantee feasibility of each of the iterates. We bound the distance from the approximate solutions to the optimal solution as a function of the variation in optimal power allocation, and we verify our results via experiments., QC 20181210

We present LShapedSolvers.jl, a suite of scalable stochastic programming solvers implemented in the Julia programming language. The solvers, which are based on the L-shaped algorithm, run efficiently in parallel, exploit problem structure, and operate on distributed data. The implementation introduces several flexible high-level abstractions that result in a modular design and simplify the development of algorithm variants. In addition, we demonstrate how the abstractions available in the Julia module for distributed computing are exploited to simplify the implementation of the parallel algorithms. The performance of the solvers is evaluated on large-scale problems for finding optimal orders on the Nordic day-ahead electricity market. With 16 worker cores, the fastest algorithm solves a distributed problem with 2.5 million variables and 1.5 million linear constraints about 19 times faster than Gurobi is able to solve the extended form directly., QC 20190423

This paper studies the synchronization of a finite number of Kuramoto oscillators in a frequency-dependent bidirectional tree network. We assume that the coupling strength of each link in each direction is equal to the product of a common coefficient and the exogenous frequency of its corresponding source oscillator. We derive a sufficient condition for the common coupling strength in order to guarantee frequency synchronization in tree networks. Moreover, we discuss the dependency of the obtained bound on both the graph structure and the way that exogenous frequencies are distributed. Further, we present an application of the obtained result by means of an event-triggered algorithm for achieving frequency synchronization in a star network assuming that the common coupling coefficient is given., QC 20190305