The Experts below are selected from a list of 192 Experts worldwide ranked by ideXlab platform
Mark Coates - One of the best experts on this subject based on the ideXlab platform.
-
Particle Filtering With Invertible Particle Flow
IEEE Transactions on Signal Processing, 2017Co-Authors: Yunpeng Li, Mark CoatesAbstract:A key challenge when designing particle filters in high-dimensional state spaces is the construction of a proposal distribution that is close to the posterior distribution. Recent advances in particle flow filters provide a promising avenue to avoid weight degeneracy; particles drawn from the prior distribution are migrated in the state space to the posterior distribution by solving partial differential equations. Numerous particle flow filters have been proposed based on different assumptions concerning the flow dynamics. Approximations are needed in the implementation of all of these filters; as a result, the particles do not exactly match a sample drawn from the desired posterior distribution. Past efforts to correct the discrepancies involve expensive calculations of importance weights. In this paper, we present new filters which incorporate deterministic particle flows into an encompassing particle filter framework. The valuable theoretical guarantees concerning particle filter performance still apply, but we can exploit the attractive performance of the particle flow methods. The filters we describe involve a computationally efficient weight update step, arising because the embedded particle flows we design possess an Invertible Mapping property. We evaluate the proposed particle flow particle filters' performance through numerical simulations of a challenging multitarget multisensor tracking scenario and complex high-dimensional filtering examples.
-
Fast particle flow particle filters via clustering
2016 19th International Conference on Information Fusion (FUSION), 2016Co-Authors: Yunpeng Li, Mark CoatesAbstract:Particle flow filters, introduced in a series of papers by Daum and Huang, are an attractive alternative to particle filters for filtering tasks in high-dimensional spaces or with very informative measurements. Many variants of particle flow filters have been developed, but all require approximations in multiple stages of the implementation, which leads to particles deviating from the true posterior distribution. To preserve the statistical consistency of the filtering algorithm, some recent papers embed the particle flow techniques within a particle filter, using them to generate a proposal distribution. In recent work, we developed such a particle flow particle filter, modifying the flow mechanism to ensure that the implemented, approximate flow was an Invertible Mapping. This property allows efficient computation of the importance weights. In this paper, we strive to reduce the computational overhead of the particle flow particle filter by incorporating clustering of the particles. Results from a multi-target acoustic tracking simulation demonstrate that we can significantly reduce the computational cost of particle flow particle filters with a relative small sacrifice in tracking accuracy.
-
FUSION - Fast particle flow particle filters via clustering
2016Co-Authors: Yunpeng Li, Mark CoatesAbstract:Particle flow filters, introduced in a series of papers by Daum and Huang, are an attractive alternative to particle filters for filtering tasks in high-dimensional spaces or with very informative measurements. Many variants of particle flow filters have been developed, but all require approximations in multiple stages of the implementation, which leads to particles deviating from the true posterior distribution. To preserve the statistical consistency of the filtering algorithm, some recent papers embed the particle flow techniques within a particle filter, using them to generate a proposal distribution. In recent work, we developed such a particle flow particle filter, modifying the flow mechanism to ensure that the implemented, approximate flow was an Invertible Mapping. This property allows efficient computation of the importance weights. In this paper, we strive to reduce the computational overhead of the particle flow particle filter by incorporating clustering of the particles. Results from a multi-target acoustic tracking simulation demonstrate that we can significantly reduce the computational cost of particle flow particle filters with a relative small sacrifice in tracking accuracy.
Yunpeng Li - One of the best experts on this subject based on the ideXlab platform.
-
Particle Filtering With Invertible Particle Flow
IEEE Transactions on Signal Processing, 2017Co-Authors: Yunpeng Li, Mark CoatesAbstract:A key challenge when designing particle filters in high-dimensional state spaces is the construction of a proposal distribution that is close to the posterior distribution. Recent advances in particle flow filters provide a promising avenue to avoid weight degeneracy; particles drawn from the prior distribution are migrated in the state space to the posterior distribution by solving partial differential equations. Numerous particle flow filters have been proposed based on different assumptions concerning the flow dynamics. Approximations are needed in the implementation of all of these filters; as a result, the particles do not exactly match a sample drawn from the desired posterior distribution. Past efforts to correct the discrepancies involve expensive calculations of importance weights. In this paper, we present new filters which incorporate deterministic particle flows into an encompassing particle filter framework. The valuable theoretical guarantees concerning particle filter performance still apply, but we can exploit the attractive performance of the particle flow methods. The filters we describe involve a computationally efficient weight update step, arising because the embedded particle flows we design possess an Invertible Mapping property. We evaluate the proposed particle flow particle filters' performance through numerical simulations of a challenging multitarget multisensor tracking scenario and complex high-dimensional filtering examples.
-
Fast particle flow particle filters via clustering
2016 19th International Conference on Information Fusion (FUSION), 2016Co-Authors: Yunpeng Li, Mark CoatesAbstract:Particle flow filters, introduced in a series of papers by Daum and Huang, are an attractive alternative to particle filters for filtering tasks in high-dimensional spaces or with very informative measurements. Many variants of particle flow filters have been developed, but all require approximations in multiple stages of the implementation, which leads to particles deviating from the true posterior distribution. To preserve the statistical consistency of the filtering algorithm, some recent papers embed the particle flow techniques within a particle filter, using them to generate a proposal distribution. In recent work, we developed such a particle flow particle filter, modifying the flow mechanism to ensure that the implemented, approximate flow was an Invertible Mapping. This property allows efficient computation of the importance weights. In this paper, we strive to reduce the computational overhead of the particle flow particle filter by incorporating clustering of the particles. Results from a multi-target acoustic tracking simulation demonstrate that we can significantly reduce the computational cost of particle flow particle filters with a relative small sacrifice in tracking accuracy.
-
FUSION - Fast particle flow particle filters via clustering
2016Co-Authors: Yunpeng Li, Mark CoatesAbstract:Particle flow filters, introduced in a series of papers by Daum and Huang, are an attractive alternative to particle filters for filtering tasks in high-dimensional spaces or with very informative measurements. Many variants of particle flow filters have been developed, but all require approximations in multiple stages of the implementation, which leads to particles deviating from the true posterior distribution. To preserve the statistical consistency of the filtering algorithm, some recent papers embed the particle flow techniques within a particle filter, using them to generate a proposal distribution. In recent work, we developed such a particle flow particle filter, modifying the flow mechanism to ensure that the implemented, approximate flow was an Invertible Mapping. This property allows efficient computation of the importance weights. In this paper, we strive to reduce the computational overhead of the particle flow particle filter by incorporating clustering of the particles. Results from a multi-target acoustic tracking simulation demonstrate that we can significantly reduce the computational cost of particle flow particle filters with a relative small sacrifice in tracking accuracy.
Pierre Moulin - One of the best experts on this subject based on the ideXlab platform.
-
Universal Decoding of Watermarks Under Geometric Attacks
2006 IEEE International Symposium on Information Theory, 2006Co-Authors: Pierre MoulinAbstract:Designing watermarking codes that can with stand geometric and other desynchronization attacks is a notoriously difficult problem. One may ask whether these difficulties are due to limitations of current codes, or rather to fundamental limitations on achievable performance. We model the attack channel as the cascade of a memoryless channel and a smooth, Invertible Mapping Tthetas, thetas isin thetasn, representing the geometric attack. The decoder does not known the value of thetas. We show that under regularity conditions, there exists a universal decoder for this problem, and we explicitly identify it
-
ISIT - Universal Decoding of Watermarks Under Geometric Attacks
2006 IEEE International Symposium on Information Theory, 2006Co-Authors: Pierre MoulinAbstract:Designing watermarking codes that can with stand geometric and other desynchronization attacks is a notoriously difficult problem. One may ask whether these difficulties are due to limitations of current codes, or rather to fundamental limitations on achievable performance. We model the attack channel as the cascade of a memoryless channel and a smooth, Invertible Mapping Tthetas, thetas isin thetasn, representing the geometric attack. The decoder does not known the value of thetas. We show that under regularity conditions, there exists a universal decoder for this problem, and we explicitly identify it
Peter Schuster - One of the best experts on this subject based on the ideXlab platform.
-
Prediction of RNA secondary structures: From theory to models and real molecules
Reports on Progress in Physics, 2006Co-Authors: Peter SchusterAbstract:RNA secondary structures are derived from RNA sequences, which are strings built form the natural\r four letter nucleotide alphabet, { AUGC }. These coarse-grained structures, in turn, are tantamount\r to constrained strings over a three letter alphabet. Hence, the secondary structures are discrete\r objects and the number of sequences always exceeds the number of structures. The sequences built\r from two letter alphabets form perfect structures when the nucleotides can form a base pair, as is\r the case with { GC } or { AU }, but the relation between the sequences and structures differs\r strongly from the four letter alphabet. A comprehensive theory of RNA structure is presented, which\r is based on the concepts of sequence space and shape space , being a space of structures. It sets\r the stage for modelling processes in ensembles of RNA molecules like evolutionary optimization or\r kinetic folding as dynamical phenomena guided by Mappings between the two spaces.\r \r The number of minimum free energy (mfe) structures is always smaller than the number of sequences,\r even for two letter alphabets. Folding of RNA molecules into mfe energy structures constitutes a\r non-Invertible Mapping from sequence space onto shape space. The preimage of a structure in sequence\r space is defined as its neutral network . Similarly the set of suboptimal structures is the preimage\r of a sequence in shape space. This set represents the conformation space of a given sequence. The\r evolutionary optimization of structures in populations is a process taking place in sequence space,\r whereas kinetic folding occurs in molecular ensembles that optimize free energy in conformation\r space. Efficient folding algorithms based on dynamic programming are available for the prediction of\r secondary structures for given sequences. The inverse problem, the computation of sequences for\r predefined structures, is an important tool for the design of RNA molecules with tailored\r properties. Simultaneous folding or cofolding of two or more RNA molecules can be modelled readily\r at the secondary structure level and allows prediction of the most stable (mfe) conformations of\r complexes together with suboptimal states. Cofolding algorithms are important tools for efficient\r and highly specific primer design in the polymerase chain reaction (PCR) and help to explain the\r mechanisms of small interference RNA (si-RNA) molecules in gene regulation.\r \r The evolutionary optimization of RNA structures is illustrated by the search for a target structure\r and mimics aptamer selection in evolutionary biotechnology. It occurs typically in steps consisting\r of short adaptive phases interrupted by long epochs of little or no obvious progress in\r optimization. During these quasi-stationary epochs the populations are essentially confined to\r neutral networks where they search for sequences that allow a continuation of the adaptive process.\r Modelling RNA evolution as a simultaneous process in sequence and shape space provides answers to\r questions of the optimal population size and mutation rates.\r \r Kinetic folding is a stochastic process in conformation space. Exact solutions are derived by direct\r simulation in the form of trajectory sampling or by solving the master equation. The exact solutions\r can be approximated straightforwardly by Arrhenius kinetics on barrier trees, which represent\r simplified versions of conformational energy landscapes. The existence of at least one sequence\r forming any arbitrarily chosen pair of structures is granted by the intersection theorem . Folding\r kinetics is the key to understanding and designing multistable RNA molecules or RNA switches . These\r RNAs form two or more long lived conformations, and conformational changes occur either\r spontaneously or are induced through binding of small molecules or other biopolymers. RNA switches\r are found in nature where they act as elements in genetic and metabolic regulation.\r \r The reliability of RNA secondary structure prediction is limited by the accuracy with which the\r empirical parameters can be determined and by principal deficiencies, for example by the lack of\r energy contributions resulting from tertiary interactions. In addition, native structures may be\r determined by folding kinetics rather than by thermodynamics. We address the first problem by\r considering base pair probabilities or base pairing entropies, which are derived from the partition\r function of conformations. A high base pair probability corresponding to a low pairing entropy is\r taken as an indicator of a high reliability of prediction. Pseudoknots are discussed as an example\r of a tertiary interaction that is highly important for RNA function. Moreover, pseudoknot formation\r is readily incorporated into structure prediction algorithms.\r \r Some examples of experimental data on RNA secondary structures that are readily explained using the\r landscape concept are presented. They deal with (i) properties of RNA molecules with random\r sequences, (ii) RNA molecules from restricted alphabets, (iii) existence of neutral networks, (iv)\r shape space covering, (v) riboswitches and (vi) evolution of non-coding RNAs as an example of\r evolution restricted to neutral networks.
Oswaldo Luiz Do Valle Costa - One of the best experts on this subject based on the ideXlab platform.
-
Stationary distributions for piecewise-deterministic Markov processes
Journal of Applied Probability, 1990Co-Authors: Oswaldo Luiz Do Valle CostaAbstract:We show that the problem of existence and uniqueness of stationary distributions for piecewise-deterministic Markov processes (PDPs) is equivalent to the same problem for the associated Markov chain, so long as some mild conditions on the parameters of the PDP are satisfied. Our main result is the construction of an Invertible Mapping from the set of stationary distributions for the PDP to the set of stationary distributions for the Markov chain. An application to capacity expansion is given
-
Stationary distributions for piecewise-deterministic Markov processes
Journal of Applied Probability, 1990Co-Authors: Oswaldo Luiz Do Valle CostaAbstract:In this paper we show that the problem of existence and uniqueness of stationary distributions for piecewise-deterministic Markov processes (PDPs) is equivalent to the same problem for the associated Markov chain, so long as some mild conditions on the parameters of the PDP are satisfied. Our main result is the construction of an Invertible Mapping from the set of stationary distributions for the PDP to the set of stationary distributions for the Markov chain. Some sufficient conditions for existence are presented and an application to capacity expansion is given.
