The scarcity of medical data, particularly in Survival Analysis (SA) for cancer-related diseases, challenges data-driven healthcare research. While Synthetic Tabular Data Generation (STDG) models have been proposed to address this issue, most rely on datasets with abundant samples, not reflecting real-world limitations. We suggest using an STDG approach that leverages transfer learning and meta-learning techniques to create an artificial inductive bias, guiding generative models trained on limited samples. Initial experiments were conducted on larger classification datasets, allowing us to asses the methodology across varying sample sizes and abundant versus scarce data scenarios. We primarily employed clinical utility validation for cancer-related SA data, as divergence-based similarity validation was not feasible. The methodology improved STDG under constrained data conditions, with divergence-based similarity validation proving to be a robust measure of data quality. Conversely, clinical utility validation yielded similar results regardless of sample size, indicating its limitations in statistically confirming effective STDG. In SA experiments, we observed that altering the task can reveal if relationships among variables are accurately generated, with most cases benefiting from the proposed methodology. Our study underscores the efficacy of the approach in tackling medical data scarcity by effectively generating high-quality synthetic data under constrained conditions. While divergence-based similarity validation is essential when sufficient data are available, clinical utility validation alone is insufficient and should be complemented by similarity validation. These findings underscore the potential and limitations of STDG methodologies in addressing medical data scarcity.

We introduce DeCaFlow, a deconfounding causal generative model. Training once per dataset using just observational data and the underlying causal graph, DeCaFlow enables accurate causal inference on continuous variables under the presence of hidden confounders. Specifically, we extend previous results on causal estimation under hidden confounding to show that a single instance of DeCaFlow provides correct estimates for all causal queries identifiable with do-calculus, leveraging proxy variables to adjust for the causal effects when do-calculus alone is insufficient. Moreover, we show that counterfactual queries are identifiable as long as their interventional counterparts are identifiable, and thus are also correctly estimated by DeCaFlow. Our empirical results on diverse settings—including the Ecoli70 dataset, with 3 independent hidden confounders, tens of observed variables and hundreds of causal queries—show that DeCaFlow outperforms existing approaches, while demonstrating its out-of-the-box applicability to any given causal graph.

While synthetic tabular data generation using Deep Generative Models (DGMs) offers a compelling solution to data scarcity and privacy concerns, their effectiveness relies on the availability of substantial training data, often lacking in real-world scenarios. To overcome this limitation, we propose a novel methodology that explicitly integrates artificial inductive biases into the generative process to improve data quality in low-data regimes. Our framework leverages transfer learning and meta-learning techniques to construct and inject informative inductive biases into DGMs. We evaluate four approaches (pre-training, model averaging, Model-Agnostic Meta-Learning (MAML), and Domain Randomized Search (DRS)) and analyze their impact on the quality of the generated text. Experimental results show that incorporating inductive bias substantially improves performance, with transfer learning methods outperforming meta-learning, achieving up to 60% gains in Jensen-Shannon divergence. The methodology is model-agnostic and especially relevant in domains such as healthcare and finance, where high-quality synthetic data are essential, and data availability is often limited.

Water Modelling (WaM) faces numerous challenges to increase its social usefulness. To boost its applicability, a broader and more robust methodological framework fis needed to face the most important WaM challenges. This research aims to provide a broad and stochastic framework, called Meta-Water-Modelling (Meta-WaM), through a surrogate model approach based on the main WaM challenges. Conceptually, this is performed through a compartmental modelling type. The Meta-WaM potential is highlighted through a detailed development of the challenge modelling “Uncertainty through a Predictive development from the Stochastic Hydrology, UPSH”; one of the seven challenges that Meta-WaM addresses. An exhaustive analytical framework on the key factors of Uncertainty, Variability and Randomness was developed. Regarding numerical results, on one hand, when a model data contains maximum uncertainty, it is recommended its analysis with 100 % chance through a causal approach (Causality). This provides an average degree of uncertainty incorporation (quality) of 36.88 %. Data set with high variability should be appropriately modelled through Multivariate approach (100 % chance), with a quality of 36.15 %. In the case of samples for modelling with high randomness, the results are not as definitive. Here the highest percentage of recommendation is in favour of the non-parametric approaches (57.14 %), with a quality of 42.03 %, in line with the data characteristics. UPSH module has been able to highlight that the randomness parameter is a crucial issue to improve hydrological behaviour. Ultimately, Meta-WaM aims to become a comprehensive stochastic decision support system to improve the applicability of water modelling developments.

Federated Learning is an emerging, powerful approach that allows training an artificial intelligence model in distributed setting. Two survival models, Cox and DeepSurv, have been trained in a federated setting, exploiting both code simulations and real experiments on the new platform, developed by the GenoMed4All consortium. Different scenarios have been tested by splitting a Myelodysplastic Syndrome dataset into three nodes and performing feature removal. A significant gain in model performance has been observed due to federated aggregation.

Discounted repeated games are currently being used to model the conflicts that arise between the nodes in a wireless network, such as distributed resource allocation, interference management or defending the network against attacks. In current literature, it is frequent that authors devise a specific strategy that performs well only for their concrete problem, thus, it would be desirable to have a generic algorithm that allows learning strategies for such games. However, current learning algorithms focus on average payoff repeated games, and we show analytically that there are important differences that prevent us from using such algorithms for discounted repeated games. In this work, we aim to fill this gap and we propose LEWIS, a lightweight, online learning algorithm specifically designed for these games, that deals with imperfect and incomplete information and is able to return a good payoff. We test LEWIS on two settings based on current literature problems to show that it has a good performance, hence, being a promising method to learn how to play a discounted repeated game in wireless networks.

The recent advances in Deep Learning have a significant impact on the security of wireless networks, such as intelligent attackers which are able to successfully exploit a possibly unknown defense mechanism simply by interacting with it. Their capacity of adapting to standard defense mechanisms, such as statistical tests, makes them a significant threat. In this article, we develop two intelligent defense mechanisms using inverse reinforcement learning tools, that can be used to enhance the capabilities of current defense mechanisms. We test our proposal on a backoff attack setup against an intelligent attacker, obtaining very significant gains in the defense performance.

Safety in coastal areas such as beaches, ports, pontoons, etc., is a current problem with a difficult solution and on which many organizations are putting efforts in terms of technological innovation. In this work the design of a possible solution based on magnetic sensors is presented. First, a study has been made of the type of sensors that best suit the application based on parameters such as sensitivity, the allowed bandwidth of excitation, price or physical construction. Then the system of excitation of the sensors and signal measurement is presented. To justify the design, a series of simulations of magnetic field variations have been carried out in the presence of large objects of conductive material, in the vicinity of the measuring points. With these data a mathematical model has been established that allows the identification of the dimensions and position of the object through triangulation and knowing only the data of the magnetic field. It was found that although this method seems quite effective, it has a significant error, so another method based on neural networks was developed also using data from the simulations. This method seems to yield much better and more reliable results.

We study an underwater navigation problem, where an Under-water Autonomous Vehicle must reach a target position in the presence of a disturbance that may be unknown. In order to deal with this problem, we make use of Deep Reinforcement Learning tools, and more concretely, we make use of robust control ideas, which allow training an agent in the presence of uncertainty. We propose a robust Proximal Policy Optimization agent and train it using simulations of an underwater medium: this agent shows an excellent performance when facing unknown disturbances, being able to approach the performance of the optimal agent which had an exact knowledge of the underwater disturbance.

In this work, we develop an algorithm based on graph networks to train distributedly a deep learning model. We consider that there are several nodes, in an arbitrary network topology, each one of them having access to a local dataset that, for privacy concerns, cannot be shared with other nodes. We also assume that there are bandwidth constraints, and hence, it may not be affordable sharing the weights of the model with other nodes. Our algorithm makes use of pruning and an autoencoder to train under all these constraints, and our simulations show that it provides a good accuracy, while preserving the privacy of the data and providing a compression of nearly two orders of magnitude in the transmitted bits.

We extend earlier works on continuous potential games to the most general case: stochastic time varying environment, stochastic rewards, non-reduced form and constrained state-action sets. We provide conditions for a Markov Nash equilibrium (MNE) of the game to be equivalent to the solution of a single control problem. Then, we address the problem of learning this MNE when the reward and state transition models are unknown. We follow a reinforcement learning approach and extend previous algorithms for working with constrained state-action subsets of real vector spaces. As an application example, we simulate a network flow optimization model, in which the relays have batteries that deplete with a random factor. The results obtained with the proposed framework are close to optimal.

We consider the problem of solving a quadratic potential game with single quadratic constraints, under no monotonicity condition of the game, nor convexity in any of the player's problem. We show existence of Nash equilibria (NE) in the game, and propose a framework to calculate Pareto efficient solutions. Regarding the corresponding non-convex potential function, we show that strong duality holds with its corresponding dual problem, give existence results of solutions and present conditions for global optimality. Finally, we propose a centralized method to solve the potential problem, and a distributed version for compact constraints. We also present simulations showing convergence behavior of the proposed distributed algorithm.

We combine model-based methods and distributed stochastic approximation to propose a fully distributed algorithm for nonconvex optimization, with good empirical performance and convergence guarantees. Neither the expression of the objective nor its gradient are known. Instead, the objective is like a “black-box”, in which the agents input candidate solutions and evaluate the output. Without central coordination, the distributed algorithm naturally balances the computational load among the agents. This is especially relevant when many samples are needed (e.g., for high-dimensional objectives) or when evaluating each sample is costly. Numerical experiments over a difficult benchmark show that the networked agents match the performance of a centralized architecture, being able to approach the global optimum, while none of the individual noncooperative agents could by itself.

Optimum scheduling is a key objective in many communications systems where different users have to share a common resource. Typically, centralized implementations are capable of guaranteeing certain fairness. In our approach, we follow a different path modeling the scheduling process as a dynamic infinite horizon discrete-time game. This formulation allows us to include any kind of dynamics and distributed implementations. Despite, these games are very difficult to solve, we are able to show that they are in fact dynamic potential games equivalent to a non-stationary multivariate optimum control problem. The dynamic control problem is solved via an augmented Bellman equation including time as an extra state.

We apply diffusion strategies to propose a cooperative reinforcement learning algorithm, in which agents in a network communicate with their neighbors to improve predictions about their environment. The algorithm is suitable to learn off-policy even in large state spaces. We provide a mean-square-error performance analysis under constant step-sizes. The gain of cooperation in the form of more stability and less bias and variance in the prediction error, is illustrated in the context of a classical model. We show that the improvement in performance is especially significant when the behavior policy of the agents is different from the target policy under evaluation.

We address a cognitive radio scenario, where a number of secondary users performs identification of which primary user, if any, is transmitting, in a distributed way and using limited location information. We propose two fully distributed algorithms: the first is a direct identification scheme, and in the other a distributed sub-optimal detection based on a simplified Neyman-Pearson energy detector precedes the identification scheme. Both algorithms are studied analytically in a realistic transmission scenario, and the advantage obtained by detection pre-processing is also verified via simulation. Finally, we give details of their fully distributed implementation via consensus averaging algorithms.

Energy efficiency is a major design issue in the context of Wireless Sensor Networks (WSN). If data is to be sent to a far-away base station, collaborative beamforming by the sensors may help to distribute the load among the nodes and reduce fast battery depletion. However, collaborative beamforming techniques are far from optimality and in many cases may be wasting more power than required. In this contribution we consider the issue of energy efficiency in beamforming applications. Using a convex optimization framework, we propose the design of a virtual beamformer that maximizes the network's lifetime while satisfying a pre-specified Quality of Service (QoS) requirement. A distributed consensus-based algorithm for the computation of the optimal beamformer is also provided.

Linear discriminant analysis (LDA) is a widely used feature extraction method for classification. We introduce distributed implementations of different versions of LDA, suitable for many real applications. Classical eigen-formulation, iterative optimization of the subspace, and regularized LDA can be asymptotically approximated by all the nodes through local computations and single-hop communications among neighbors. These methods are based on the computation of the scatter matrices, so we introduce how to estimate them in a distributed fashion. We test the algorithms in a realistic distributed classification problem, achieving a performance near to the centralized solution and a significant improvement of 35% over the non-cooperative case.

Nonparametric belief propagation (NBP) is a well-known particle-based method for distributed inference in wireless networks. NBP has a large number of applications, including cooperative localization. However, in loopy networks NBP suffers from similar problems as standard BP, such as over-confident beliefs and possible non-convergence. Tree-reweighted NBP (TRW-NBP) can mitigate these problems, but does not easily lead to a distributed implementation due to the non-local nature of the required so-called edge appearance probabilities. In this paper, we propose a variation of TRW-NBP, suitable for cooperative localization in wireless networks. Our algorithm uses a fixed edge appearance probability for every edge, and can outperform standard NBP in dense wireless networks.

Study and evaluation of wireless sensor networks (WSN) require modeling of different elements and settling performance boundaries. Available IEEE 802.15.4 models make their own assumptions of the standard within a given scenario. Such assumptions (and interpretations) do not really match the performance of real equipment (motes) sometimes. Here we propose a model based on one of the most utilized WSN platforms: micaZ. Its operation and timings are described as well so that they can be used for posterior modeling if other scenarios need to be studied. The throughput upper bound of peer-to-peer micaZ-TinyOS links is provided based both on our model and real measurements for the full payload range. The proposed model is compared to both real measurements and a model based on the standard. Results show our model to be accurate and better matching reality for any payload.

Nonparametric belief propagation (NBP) is one of the best-known methods for cooperative localization in sensor networks. It is capable to provide information about location estimation with appropriate uncertainty and to accommodate non-Gaussian distance measurement errors. However, the accuracy of NBP is questionable in loopy networks. In this paper, we propose a novel approach, NBP based on spanning trees (NBP-ST) created by breadth first search (BFS) method. According to our simulation results, NBP-ST performs better than NBP in terms of accuracy, computational and communication cost in the networks with high connectivity (i.e., highly loopy networks).

This paper investigates common-mode propagation in shielded twisted pair cables. The common mode exhibits great potential for improving the throughput in emerging wireline systems. The design of corresponding transmission schemes over multipair copper cables requires accurate knowledge of the channel properties. We present measurement and modeling results and investigate the feasibility of using standard differential- mode models for data fitting in multiconductor transmission-line modelling of common-mode paths.

We compare the performance between beamforming and spatial multiplexing showing in which downlink scenarios the higher performance of spatial multiplexing justify its complexity. We compute performance using readily measurable parameters such as angle spread (AS), antenna separation and signal to noise ratio (SKIR). Firstly, a semi-analytical approach relates these measurable parameters with parameters that theoretically characterize beamforming optimality such as the spatial correlation matrix first two eigenvalues and SNR. Secondly, the achieved spectral efficiency is given for beamforming and spatial multiplexing as a function of antenna separation, AS and SNR. Also, a "practical" region is given where beamforming achieves at least 90% of the spectral efficiency of spatial multiplexing.

We consider the MIMO broadcast channel where a base station with multiple antennas transmits simultaneously to many users. The optimum strategy with channel state information (CSI) at the transmitter is based on dirty paper coding (DPC) principles. In this paper, we propose a suboptimum strategy suitable when terminals are equipped with multiple antennas. This approach combines DPC and receive beamforming and is shown to provide near-optimum performance with very reasonable complexity. Beamforming design is based on the maximization of the sum rate while precoder design is based on zero forcing (ZF) criteria. We have to remark that it is not required iterative processes or interaction between transmitter and receivers to find the suboptimum solution. It is also important to mention that each terminal just needs to know its own channel to perform the optimization while the base station requires full knowledge of all users' channels

Modified digital prolate functions (DPF) are addressed in this paper, applied as waveforms for downlink MC-CDMA systems and providing higher transmission rates. Those waveforms, thanks to their double orthogonality and energy concentration in restricted interval properties, enable specific processing at the receiver. Design and analysis of advanced receivers exploiting the abovementioned DPF properties and computer simulations that demonstrate their performance improvement compared to MC-CDMA system with Hadamard codes are presented.

Multiple Access Interference (MAI) imposes serious limitation to the MC-CDMA systems' performance in uplink channel. In our proposal we are considering a system where orthogonal users' codes are based on Prolate Spheroidal Functions (PSF), specifically those denominated Periodic Discrete Prolate Spheroidal Sequences (PDPSS). These codes have the property concentrating signal energy in the desired time interval and gives the possibility to avoid Inter Symbol Interference (ISI) for a certain number of users without introducing the guard interval between consecutive symbols. Furthermore, it is also possible to avoid completely MAI for a certain number of active users in the uplink channel. Even in the case when this limit is passed, MAI is below the standard MC-CDMA schemes. Preliminary simulations presented here demonstrate this affirmation.

We investigate the performance and the effects of the decorrelation process in the multi-user stage detection with several user's signals, using multiple carrier code division multiple access (MC-CDMA) modulation. We are dealing with the reverse link, assuming an asynchronous detection at the base station.

Requests for higher data transmissions rates and system performance in multiuser radio communications, needed for the future mobile generations or wireless local area networks, have centered them in the focus of interest of investigations in the signal processing area. MC-CDMA systems with orthogonal user codes based on Hadamard codes do not have any control of the transmitted waveform time spreading. In this case, a time guard is necessary to avoid the ISI effect. We have designed an alternative set of non-binary codes based on the time-concentration capability of prolate spheroidal functions. This property allows the use of several waveforms with almost ISI-free behavior. The advantage of using discrete prolate functions (DPF), instead of Hadamard codes, as users' codes is obvious; when the system is not fully loaded, users are able to transfer greater bit rates with better error performances. In a multiuser scenario, the receiver design is almost identical to the standard case. Exploiting also the DPF property of double orthogonality, implementation of a new set of matched filters in a shorter time interval is investigated.

Blind source separation problem whose solution is vital in numerous applications in communications. We are proposing a multistage procedure to separate N original sources from N instantaneous mixtures. The goal is to extract the parameters of the unknown mixture in a deflation approach. In each stage of the procedure a novel cost function is applied. The cost function is derived from the properties of the cdf (cumulative density function) to perform an appropriate independent measure by means of order statistics (OS) (unbiased estimator of the cdf). The key-point of this contribution is the adaptive algorithm applied to optimize our cost function using gradient descent techniques.

It is well known that blind channel deconvolution enables the receiver to equalize the channel simply by analyzing the received digital signal. Much of the work in 1990s faces the challenge presented by multiple-output systems, exploiting cyclostationarity properties and multivariate formulation of the incoming data. Our proposal is twofold: on one hand, we develop a theoretical analysis of a new blind channel deconvolution scheme by the exploitation of some shifting properties of the autocorrelation matrices of the source, in order to propose an appropriate cost function; on the other hand, an efficient programming is considered based on a generalized Rayleigh quotient formulation by using a conjugate gradient algorithm.

Recent works have presented novel techniques that exploit cyclostationarity for channel identification (equalization) in data communication systems using only second order statistics. In particular, the feasibility of blind identification based on the 'forward shift' structure of the correlation matrices of the source has been shown. In this paper we propose an alternative algorithm based on this property but with an improved choice of the autocorrelation matrices to be considered. This new representation of the equalization problem provide a cost function formulated as a generalized Rayleigh quotient, which may be efficiently implemented by using conjugate gradient techniques. Several simulations over different data transmission constellations support our theoretical analysis.

The objective of any local echo canceller connected to any end of a digital subscriber line (DSL) is to compensate the outgoing transmit signal energy in the receiving path (commonly called near-echo) that the hybrid circuit leaks. The observed signals that usually play the role of error references for adaptive echo cancellers (EC) in data communications are basically composed by far-end data plus a residual echo. The EC goal is to eliminate that residual echo. However, depending on the convergence stage, the presence of data in the error reference may mask the effective error reference (the residual echo). We show a new adaptive EC that offers better convergence properties than those associated to the simplest EC algorithms (standard LMS and decision-directed) for multilevel PAM signals. It exploits the discrete nature of the far-end signal to self-cancel the masking effect by adding (or subtracting) a weighted old version of the error reference to the current one. We show that the new error cost function is now closer to the instantaneous optimal one: its corresponding MLE. This new error surface presents a more robust (or uniform) behavior during the convergence period.

Existing blind adaptive equalizers that use nonconvex cost functions (as Bussgang type algorithms) and stochastic gradient descent suffer from lack of global convergence to an equalizer tap set that removes sufficient ISI when an FIR equalizer is used. In this paper we propose a new algorithm including tap anchoring and gain recovery into the classical schemes. The combined effect of these strategies is to establish the preservation of the transmitted symbol preventing ill convergence, and therefore providing the ability of implementation of the inverse filter regardless of the initial ISI. Under certain hypotheses, we suggest that a globally convex scheme can be proposed overcoming the existing structures. Several computer simulations support our theoretical results.

The observed signals that usually play the role of error references for adaptive echo cancellers and channel equalizers in data communications are basically composed by data plus a residual error. The goal of such systems is to eliminate those errors to enhance the present data. However, in some systems, depending on the convergence stage, the presence of data in the error reference may mask or distort the true error reference (the residual echo or ISI). This paper shows a new perspective on how to use old samples of the observed error to improve the current error reference. It exploits the discrete or the constant envelope nature of the data to build up an error cost function that matches the instantaneous optimal one: their corresponding MLE. These new error surfaces present a more robust (or uniform) behavior during the convergence period since they offer a proper unmasked residual error as error reference.>

Presents a new adaptive algorithm for joint blind equalization and phase recovery in data communication systems. This algorithm is mainly based on the standard decision directed algorithm (DDA) but the incorporation of both a linear constraint in the filter coefficients and an adjustable complex parameter in the decision device, provide an unimodal error surface. This attractive performance is analyzed theoretically when faced with minimum phase channels. For the more realistic situation of non minimum phase channels exhaustive simulations show that a similar behavior can be expected.>