We consider the query recommendation problem in closed loop interactive learning settings like online information gathering and exploratory analytics. The problem can be naturally modelled using the Multi-Armed Bandits (MAB) framework with countably many arms. The standard MAB algorithms for countably many arms begin with selecting a random set of candidate arms and then applying standard MAB algorithms, e.g., UCB, on this candidate set downstream. We show that such a selection strategy often results in higher cumulative regret and to this end, we propose a selection strategy based on the maximum utility of the arms. We show that in tasks like online information gathering, where sequential query recommendations are employed, the sequences of queries are correlated and the number of potentially optimal queries can be reduced to a manageable size by selecting queries with maximum utility with respect to the currently executing query. Our experimental results using a recent real online literature discovery service log file demonstrate that the proposed arm selection strategy improves the cumulative regret substantially with respect to the state-of-the-art baseline algorithms. Our data model and source code are available at \url{https://anonymous.4open.science/r/0e5ad6b7-ac02-4577-9212-c9d505d3dbdb/}

In spite of the overwhelming high-tech marvels and applications that rule our digital lives, the use of the handwritten signature is still recognized worldwide in government, personal and legal entities to be the most important behavioral biometric trait. A number of notable research approaches provide advanced results up to a certain point which allow us to assert with confidence that the performance attained by signature verification (SV) systems is comparable to those provided by any other biometric modality. Up to now, the mainstream trend for offline SV is shared between standard -or handcrafted- feature extraction methods and popular machine learning techniques, with typical examples ranging from sparse representation to Deep Learning. Recent progress in graph mining algorithms provide us with the prospect to re-evaluate the opportunity of utilizing graph representations by exploring corresponding graph features for offline SV. In this paper, inspired by the recent use of image visibility graphs for mapping images into networks, we introduce for the first time in offline SV literature their use as a parameter free, agnostic representation for exploring global as well as local information. Global properties of the sparsely located content of the shape of the signature image are encoded with topological information of the whole graph. In addition, local pixel patches are encoded by sequential visibility motifs-subgraphs of size four, to a low six dimensional motif profile vector. A number of pooling functions operate on the motif codes in a spatial pyramid context in order to create the final feature vector. The effectiveness of the proposed method is evaluated with the use of two popular datasets. The local visibility graph features are considered to be highly informative for SV; this is sustained by the corresponding results which are at least comparable with other classic state-of-the-art approaches.

We propose a cross-layer scheme for regulating power consumption in energy-constrained ad hoc wireless networks where information is disseminated in an adaptive epidemic manner. A time-optimized mechanism based on the Optimal Stopping Theory utilizes noise fluctuations to trigger the adaptation of transmission characteristics in a fashion that a perceived net reward is maximized. This results in data delivery being enhanced while energy cost remains modest.

This work introduces a novel representation for the off-line handwritten signature by measuring its structural characteristics with a set of partially ordered grid features. Thirty-two binary symbols are delineated within the five-by-five pixel window and considered to be the alphabet of a probabilistic source. Subsequently and for practical purposes the whole set is appropriately organized into subsets of four symbols each. The new arrangement is used to detect the presence of simple or compound symbols in the signature image. The utilization of the partially ordered set or poset notion, intuitively arranges the binary feature extraction masks into first order chains, creating in this way a first order probabilistic description of the signature's structure that is characteristic of the motoric signature generating process. In order to estimate these probabilities, the first order searching strategy is limited to pixels neighbors having their grids centered to a predetermined Chebyshev distance of two. The utilized verification strategy relies on an SVM based classifier and the equal error rate figure. Our experimental scheme handles skilled forgery as well and derives verification results for two datasets, the GPDS300 and a proprietary one.

Contextual data collection is a major challenge in mobile sensor networks, in which sources generate quality-stamped context and mobile nodes (collectors) attempt to gather context of high quality. We deal with the context collection problem, in which collectors forage for high quality context and, then, deliver it to mobile context-aware applications. Collectors undergo a context collection process by exchanging contextual data with neighbouring collectors and/or sources in light of receiving context of better quality. The quality indicator of context normally decreases with time. Hence, the collectors cannot prolong such process forever, where the delivered context might be useless for the application. We propose a context collection scheme based on the optimal stopping theory (OST), which supports collectors with time-optimized context delivery decisions. We compare the performance of the proposed scheme with pre-existing OST-based context collection mechanism and quantify the benefits stemming for its adoption.

Ad hoc networks such as Wireless Sensor Networks (WSN) are characterized by the scarcity of energy resources (among other resource types). Their intelligent design can extend their lifetime without compromising the operation of network nodes and the applications running on them. To this end we propose an adaptive epidemic scheme that helps reduce energy expenditure through intelligent tuning of the infection (forwarding) rate. Redundant communications and experienced error rate drive the adaptation of the forwarding rate. Simulation results show that significant energy gains can be obtained through the proposed scheme.

In our everyday life, we frequently experience cases where persons group together. In those cases, context-aware systems capture and process identical context. Therefore, the need to collaboratively address Context Awareness (CA) emerges. In the considered setting, ad hoc networking between mobile nodes enables the exchange of information, thus, CA is facilitated. The synergy between mobile nodes materializes the Collaborative CA (CCA) paradigm. We advance the general CCA concept by performing communication between nodes probabilistically in a way similar to virus (epidemic) spreading. Nodes feature a hierarchical information model, which can be exploited by an information diffusion process. Multiple pieces of information, exchanged as epidemics, can complete the information present at a certain node, which in turn infers and spreads new information. We study this novel scheme extensively through an information model for context and an analytical framework (Markov process) with simulations. Our findings show that the information spreading large benefits from the mobility of nodes and semantic processing of the information model.

Fires are a common, disastrous phenomenon that constitutes a serious threat. Thus, early detection is of great importance as the consequences of a fire are catastrophic. Towards this direction the SCIER project envisages the deployment of wireless sensor networks at the "urban-rural-interface" (URI) and uses sensor fusion techniques to enhance the performance of the early fire detection and fire location estimation processes.

In this paper we discuss a system that exploits observations derived from sensors, in order to estimate a key factor for pervasive computing and context aware applications: the location of a user. The term "sensors" includes Wi-Fi adapters, IR receivers, etc. The core of the system is the fusion engine which is based on dynamic Bayesian networks (DBNs), a powerful mathematical tool for integrating heterogeneous sensor observations. In closing, is provided an evaluation of the system as it comes out from the experimental results

A flow rejection algorithm for the dynamic resource reservation protocol (dRSVP) protocol is proposed. dRSVP is an enhanced version of RSVP, aiming at providing dynamic quality-of-service (QoS) support in a variable bandwidth environment (e.g., wireless) through guarantees of ranges of bandwidth instead of specific values. Flow rejection occurs when the channel quality decreases to a level that even the minimum bandwidth requirements per flow cannot be fulfilled. The proposed algorithm aims at improving the flow dropping probability of dRSVP, without affecting the bandwidth utilization. Its operation is based on setting priority classes and rejecting the minimum required number of flows per class, in order to guarantee QoS to the remaining. Both mathematical analysis and simulation results show that the overall flow dropping probability can be significantly reduced.

A flow rejection algorithm for the dynamic RSVP (dRSVP) is proposed. dRSVP is an enhanced version of RSVP, aiming at providing dynamic quality of service (QoS) support in a variable bandwidth environment through guarantees of ranges of bandwidth instead of specific values. Flow rejection occurs when the channel quality decreases to a level that even the minimum bandwidth requirements per flow cannot be fulfilled. The proposed algorithm aims at significantly improving the flow dropping probability, without affecting the bandwidth utilization. Its operation is based on setting priority classes and rejecting the minimum required number of flows per class, in order to guarantee QoS to the remaining. Both mathematical analysis and simulation results show that the overall flow rejection probability can be significantly reduced.

Mobility and QoS provision are the two most important burdens to solve before IP becomes truly ubiquitous. Mobile IP and RSVP (resource reservation protocol) are the most visible proposals for those problems. Their interoperability, though, is inefficient. To ameliorate the compatibility problem and performance gap, we have proposed the introduction of the RSVP mobility proxy (see Paskalis, S. et al., "RSVP Mobility Proxy", Internet Draft, 2001; Workshop on Wireless Local Networks, 2001). We present a performance evaluation of our scheme versus the plain RSVP operation.

The paper focuses on the integration of Internet and wireless LANs (WLANs) and proposes a novel technique that supports seamless and efficient integration. This technique is based on a generic adaptation layer, referred to as wireless adaptation layer (WAL), which is introduced between the IP layer and the WLAN link layer. WAL is designed to: (i) cope dynamically with wireless channel impairments; (ii) provide QoS interworking between the IP infrastructure and the WLAN; (iii) be independent of the WLAN technology. These characteristics account for the advantages of WAL, namely, improved performance, wide applicability and provisioning of differentiated services.

Mobile hosts utilize Mobile IP to retain connectivity, while roaming in various networks, and acquire new IP addresses through the mechanisms that Mobile IP supports. One of today's most demanding application requirements is QoS support. RSVP (resource reservation protocol), the protocol implementation of the IETF integrated services architecture (IntServ), cannot handle hosts that change their IP addresses during a connection lifetime, and thus, must re-establish any given reservation states. To overcome this limitation, we propose the adoption of a mobility management scheme, that maintains a single contact IP address throughout the mobility session. Furthermore, we introduce RSVP mobility proxy, an enhanced RSVP enabled border router to deal with QoS signaling at any mobility related network topology modification.

The medium access control (MAC) protocol and the underlying traffic scheduling mechanism for the radio interface are critical components of a wireless ATM network, aiming to provide ATM-like quality of service (QoS) guarantees over the air. A traffic scheduling algorithm for a wireless ATM customer premises network is proposed. The algorithm operates in the mobile terminals, and complements a centralized scheduling mechanism operating in the base stations, for improved performance. The proposed algorithm is delay oriented to meet the requirements of the various traffic classes defined by the ATM architecture. The system performance improvement attained by the proposed algorithm is evaluated via simulations.

Nonlinear adaptive filters based on a variety of neural network models have been used successfully for system identification and noise-cancellation in a wide class of applications. An important problem in data communications is that of channel equalization, i.e., the removal of interferences introduced by linear or nonlinear message corrupting mechanisms, so that the originally transmitted symbols can be recovered correctly at the receiver. In this paper we introduce an adaptive recurrent neural network (RNN) based equalizer whose small size and high performance makes it suitable for high-speed channel equalization. We propose RNN based structures for both trained adaptation and blind equalization, and we evaluate their performance via extensive simulations for a variety of signal modulations and communication channel models. It is shown that the RNN equalizers have comparable performance with traditional linear filter based equalizers when the channel interferences are relatively mild, and that they outperform them by several orders of magnitude when either the channel's transfer function has spectral nulls or severe nonlinear distortion is present. In addition, the small-size RNN equalizers, being essentially generalized IIR filters, are shown to outperform multilayer perceptron equalizers of larger computational complexity in linear and nonlinear channel equalization cases.