David Luengo | Universidad Politécnica de Madrid (original) (raw)
Papers by David Luengo
... Caos Determinista (I) Caos [RAE1992]: Estado amorfo e indefinido que se supone anterior a l... more ... Caos Determinista (I) Caos [RAE1992]: Estado amorfo e indefinido que se supone anterior a la construcci ´on del cosmos. En sentido figurado, confusi ´on y desorden. Caos determinista: Comportamiento estoc ´astico que ocurre en un sistema determinista [Stewar2001]. ...
... Madrid e-mail: luengod@ieee.org Ignacio Santamarıa Caballero Departamento de Ingenierıa de Co... more ... Madrid e-mail: luengod@ieee.org Ignacio Santamarıa Caballero Departamento de Ingenierıa de Comunicaciones Universidad de Cantabria e-mail: nacho@gtas. dicom.unican.es AbstractThe broadband nature and noise ...
Interest in multioutput kernel methods is increasing, whether under the guise of multitask learni... more Interest in multioutput kernel methods is increasing, whether under the guise of multitask learning, multisensor networks or structured output data. From the Gaussian process perspective a multioutput Mercer kernel is a covariance function over correlated output functions. One way of constructing such kernels is based on convolution processes (CP). A key problem for this approach is efficient inference. Alvarez and
Monitoring large hydroelectric rotating machines requires a multichannel system. Simultaneous sam... more Monitoring large hydroelectric rotating machines requires a multichannel system. Simultaneous sampling of several input channels is needed to avoid the propagation of initial mea- surement errors. However, this is usually not guaranteed by the acquisition hardware, due to the additional expense involved. The alternative is to correct the error caused by non-simultane ous sampling using digital signal processing tech- niques.
Interest in multioutput kernel methods is increas- ing, whether under the guise of multitask lear... more Interest in multioutput kernel methods is increas- ing, whether under the guise of multitask learn- ing, multisensor networks or structured output data. From the Gaussian process perspective a multioutput Mercer kernel is a covariance func- tion over correlated output functions. One way of constructing such kernels is based on convolution processes (CP). A key problem for this approach is efficient
We consider the problem of sequential prediction of real-valued sequences using piecewise linear ... more We consider the problem of sequential prediction of real-valued sequences using piecewise linear models under the square-error loss function. In this context, we demonstrate a sequential algorithm for prediction whose accumulated squared error for every bounded sequence is asymptotically as small as that of the best fixed predictor for that sequence taken from the class of piecewise linear predictors. We also show that this predictor is optimal in certain settings in a particular min-max sense. This approach can also be applied to the class of piecewise constant predictors, for which a similar universal sequential algorithm can be derived with corresponding min-max optimality.
IEEE Signal Processing Letters, 2015
Multiple importance sampling (MIS) methods use a set of proposal distributions from which samples... more Multiple importance sampling (MIS) methods use a set of proposal distributions from which samples are drawn. Each sample is then assigned an importance weight that can be obtained according to different strategies. This work is motivated by the trade-off between variance reduction and computational complexity of the different approaches (classical vs. deterministic mixture) available for the weight calculation. A new method that achieves an efficient compromise between both factors is introduced in this paper. It is based on forming a partition of the set of proposal distributions and computing the weights accordingly. Computer simulations show the excellent performance of the associated partial deterministic mixture MIS estimator.
2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2014
Adaptive Rejection Metropolis Sampling (ARMS) is a wellknown MCMC scheme for generating samples f... more Adaptive Rejection Metropolis Sampling (ARMS) is a wellknown MCMC scheme for generating samples from onedimensional target distributions. ARMS is widely used within Gibbs sampling, where automatic and fast samplers are often needed to draw from univariate full-conditional densities. In this work, we propose an alternative adaptive algorithm (IA 2 RMS) that overcomes the main drawback of ARMS (an uncomplete adaptation of the proposal in some cases), speeding up the convergence of the chain to the target. Numerical results show that IA 2 RMS outperforms the standard ARMS, providing a correlation among samples close to zero.
This paper presents a specific DSP based acquisition system developed for monitoring the behavior... more This paper presents a specific DSP based acquisition system developed for monitoring the behavior of low speed hydro-generator sets. The designed unit can acquire up to 16 channels and is based on the TMS320C30 DSP. It implements several digital signal processing algorithms: decimation, time delay correction to get simultaneous sampling, spectral estimation, etc. The acquired time signals, their spectra, and
IEEE Transactions on Circuits and Systems Ii: Analog and Digital Signal Processing, 2000
Digital filtering is a common approach to achieve simultaneous sampling of several input signals ... more Digital filtering is a common approach to achieve simultaneous sampling of several input signals acquired with a multiplexing delay. In this work, an error bound is obtained for Lagrange interpolation filters as a function of the oversampling ratio of the input signals, the fractional delay, and the filter's order. This bound can be used to ensure that the error is
K Traditionally, the main focus in machine learning has been model generation through a data driv... more K Traditionally, the main focus in machine learning has been model generation through a data driven paradigm.
Monte Carlo (MC) methods are well-known computational techniques, widely used in different fields... more Monte Carlo (MC) methods are well-known computational techniques, widely used in different fields such as signal processing, communications and machine learning. An important class of MC methods is composed of importance sampling (IS) and its adaptive extensions, such as population Monte Carlo (PMC) and adaptive multiple IS (AMIS). In this work, we introduce a novel adaptive and iterated importance sampler using a population of proposal densities. The proposed algorithm, named adaptive population importance sampling (APIS), provides a global estimation of the variables of interest iteratively, making use of all the samples previously generated. APIS combines a sophisticated scheme to build the IS estimators (based on the deterministic mixture approach) with a simple temporal adaptation (based on epochs). In this way, APIS is able to keep all the advantages of both AMIS and PMC, while minimizing their drawbacks. Furthermore, APIS is easily parallelizable. The cloud of proposals is adapted in such a way that local features of the target density can be better taken into account compared to single global adaptation procedures. The result is a fast, simple, robust and high-performance algorithm applicable to a wide range of problems. Numerical results show the advantages of the proposed sampling scheme in four synthetic examples and a localization problem in a wireless sensor network.
... Caos Determinista (I) Caos [RAE1992]: Estado amorfo e indefinido que se supone anterior a l... more ... Caos Determinista (I) Caos [RAE1992]: Estado amorfo e indefinido que se supone anterior a la construcci ´on del cosmos. En sentido figurado, confusi ´on y desorden. Caos determinista: Comportamiento estoc ´astico que ocurre en un sistema determinista [Stewar2001]. ...
... Madrid e-mail: luengod@ieee.org Ignacio Santamarıa Caballero Departamento de Ingenierıa de Co... more ... Madrid e-mail: luengod@ieee.org Ignacio Santamarıa Caballero Departamento de Ingenierıa de Comunicaciones Universidad de Cantabria e-mail: nacho@gtas. dicom.unican.es AbstractThe broadband nature and noise ...
Interest in multioutput kernel methods is increasing, whether under the guise of multitask learni... more Interest in multioutput kernel methods is increasing, whether under the guise of multitask learning, multisensor networks or structured output data. From the Gaussian process perspective a multioutput Mercer kernel is a covariance function over correlated output functions. One way of constructing such kernels is based on convolution processes (CP). A key problem for this approach is efficient inference. Alvarez and
Monitoring large hydroelectric rotating machines requires a multichannel system. Simultaneous sam... more Monitoring large hydroelectric rotating machines requires a multichannel system. Simultaneous sampling of several input channels is needed to avoid the propagation of initial mea- surement errors. However, this is usually not guaranteed by the acquisition hardware, due to the additional expense involved. The alternative is to correct the error caused by non-simultane ous sampling using digital signal processing tech- niques.
Interest in multioutput kernel methods is increas- ing, whether under the guise of multitask lear... more Interest in multioutput kernel methods is increas- ing, whether under the guise of multitask learn- ing, multisensor networks or structured output data. From the Gaussian process perspective a multioutput Mercer kernel is a covariance func- tion over correlated output functions. One way of constructing such kernels is based on convolution processes (CP). A key problem for this approach is efficient
We consider the problem of sequential prediction of real-valued sequences using piecewise linear ... more We consider the problem of sequential prediction of real-valued sequences using piecewise linear models under the square-error loss function. In this context, we demonstrate a sequential algorithm for prediction whose accumulated squared error for every bounded sequence is asymptotically as small as that of the best fixed predictor for that sequence taken from the class of piecewise linear predictors. We also show that this predictor is optimal in certain settings in a particular min-max sense. This approach can also be applied to the class of piecewise constant predictors, for which a similar universal sequential algorithm can be derived with corresponding min-max optimality.
IEEE Signal Processing Letters, 2015
Multiple importance sampling (MIS) methods use a set of proposal distributions from which samples... more Multiple importance sampling (MIS) methods use a set of proposal distributions from which samples are drawn. Each sample is then assigned an importance weight that can be obtained according to different strategies. This work is motivated by the trade-off between variance reduction and computational complexity of the different approaches (classical vs. deterministic mixture) available for the weight calculation. A new method that achieves an efficient compromise between both factors is introduced in this paper. It is based on forming a partition of the set of proposal distributions and computing the weights accordingly. Computer simulations show the excellent performance of the associated partial deterministic mixture MIS estimator.
2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2014
Adaptive Rejection Metropolis Sampling (ARMS) is a wellknown MCMC scheme for generating samples f... more Adaptive Rejection Metropolis Sampling (ARMS) is a wellknown MCMC scheme for generating samples from onedimensional target distributions. ARMS is widely used within Gibbs sampling, where automatic and fast samplers are often needed to draw from univariate full-conditional densities. In this work, we propose an alternative adaptive algorithm (IA 2 RMS) that overcomes the main drawback of ARMS (an uncomplete adaptation of the proposal in some cases), speeding up the convergence of the chain to the target. Numerical results show that IA 2 RMS outperforms the standard ARMS, providing a correlation among samples close to zero.
This paper presents a specific DSP based acquisition system developed for monitoring the behavior... more This paper presents a specific DSP based acquisition system developed for monitoring the behavior of low speed hydro-generator sets. The designed unit can acquire up to 16 channels and is based on the TMS320C30 DSP. It implements several digital signal processing algorithms: decimation, time delay correction to get simultaneous sampling, spectral estimation, etc. The acquired time signals, their spectra, and
IEEE Transactions on Circuits and Systems Ii: Analog and Digital Signal Processing, 2000
Digital filtering is a common approach to achieve simultaneous sampling of several input signals ... more Digital filtering is a common approach to achieve simultaneous sampling of several input signals acquired with a multiplexing delay. In this work, an error bound is obtained for Lagrange interpolation filters as a function of the oversampling ratio of the input signals, the fractional delay, and the filter's order. This bound can be used to ensure that the error is
K Traditionally, the main focus in machine learning has been model generation through a data driv... more K Traditionally, the main focus in machine learning has been model generation through a data driven paradigm.
Monte Carlo (MC) methods are well-known computational techniques, widely used in different fields... more Monte Carlo (MC) methods are well-known computational techniques, widely used in different fields such as signal processing, communications and machine learning. An important class of MC methods is composed of importance sampling (IS) and its adaptive extensions, such as population Monte Carlo (PMC) and adaptive multiple IS (AMIS). In this work, we introduce a novel adaptive and iterated importance sampler using a population of proposal densities. The proposed algorithm, named adaptive population importance sampling (APIS), provides a global estimation of the variables of interest iteratively, making use of all the samples previously generated. APIS combines a sophisticated scheme to build the IS estimators (based on the deterministic mixture approach) with a simple temporal adaptation (based on epochs). In this way, APIS is able to keep all the advantages of both AMIS and PMC, while minimizing their drawbacks. Furthermore, APIS is easily parallelizable. The cloud of proposals is adapted in such a way that local features of the target density can be better taken into account compared to single global adaptation procedures. The result is a fast, simple, robust and high-performance algorithm applicable to a wide range of problems. Numerical results show the advantages of the proposed sampling scheme in four synthetic examples and a localization problem in a wireless sensor network.