Search

Purpose To use a deep neural network (DNN) for solving the optimization problem of water/fat separation and to compare supervised and unsupervised training. Methods The current T 2 ∗ -IDEAL algorithm for solving water/fat separation is dependent on initialization. Recently, DNN has been proposed to solve water/fat separation without the need for suitable initialization. However, this approach requires supervised training of DNN using the reference water/fat separation images. Here we propose 2 novel DNN water/fat separation methods: 1) unsupervised training of DNN (UTD) using the physical forward problem as the cost function during training, and 2) no training of DNN using physical cost and backpropagation to directly reconstruct a single dataset. The supervised training of DNN, unsupervised training of DNN, and no training of DNN methods were compared with the reference T 2 ∗ -IDEAL. Results All DNN methods generated consistent water/fat separation results that agreed well with T 2 ∗ -IDEAL under proper initialization. Conclusion The water/fat separation problem can be solved using unsupervised deep neural networks.

Computing the nucleolus is recognized as an equitable solution to cooperative n person cost games, such as a vehicle routing game (VRG). Computing the nucleolus of a VRG, however, has been limited to small-sized benchmark instances with no more than 25 players, because of the computation time required to solve the NP-hard separation problem. To reduce computation time, we develop an enumerative algorithm that computes the nucleolus of the VRG with time windows (VRGTW) in the case of the non-empty core. Numerical simulations demonstrate the ability of the proposed algorithm to compute the nucleolus of benchmark instances with up to 100 players.

Scientific literature contains large volumes of complex, unstructured figures that are compound in nature (i.e. composed of multiple images, graphs, and drawings). Separation of these compound figures is critical for information retrieval from these figures. In this paper, we propose a new strategy for compound figure separation, which decomposes the compound figures into constituent subfigures while preserving the association between the subfigures and their respective caption components. We propose a two-stage framework to address the proposed compound figure separation problem. In particular, the subfigure label detection module detects all subfigure labels in the first stage. Then, in the subfigure detection module, the detected subfigure labels help to detect the subfigures by optimizing the feature selection process and providing the global layout information as extra features. Extensive experiments are conducted to validate the effectiveness and superiority of the proposed framework, which improves the detection precision by 9%.

The separation of particles from liquids represents a cross-sectional technology, which touches nearly every industrial process, our personal life and the environment. It is obvious, that very different separation principles and a huge variety of highly specialized apparatuses are needed to solve all separation problems in such different areas of application. Looking to the history of apparatus development one can observe several interesting analogies to evolutionary processes in biology. Motivation for development of new separation technology normally is a preferably urgent need to solve a separation problem and the pressure to be successful on the market. To be successful a new concept must offer advantages in comparison to competing solutions. In contrast to nature the creative “genius” of the engineer in combination with advanced knowledge makes new developments more targeted and faster. Some evolutionary aspects of separation technology will be discussed in the example of filters and centrifuges. According to actual trends latest developments are demonstrated and some future prospects are pointed out.

Part 1: Machine Learning; International audience; Linear discriminant analysis (LDA) is one of the most classical linear projection techniques for feature extraction, widely used in kinds of fields. Classical LDA is contributed to finding an optimal projection subspace that can maximize the between-class scatter and minimize the average within-class scatter of each class. However, the class separation problem always exists and classical LDA can not guarantee that the within-class scatter of each class get its minimum. In this paper, we proposed the k-classifiers method, which can reduce every within-class scatter of classes respectively and alleviate the class separation problem. This method will be applied in LDA and Norm LDA and achieve significant improvement. Extensive experiments performed on MNIST data sets demonstrate the effectiveness of k-classifiers.

In this paper, the problem of solving mixed integer bi-level programming problems is discussed. The proposed algorithm is based on the multi-parametric algorithm that using bisection cutting plane techniques to improve bounds. In fact the paper examines the nature of the so-called separation problem, which is generating a valid inequality arising as the solution to the current sub problem of the multi-parametric algorithm. Some examples are provided to demonstrate the algorithm and evaluate its performance.

Accurate estimation of the glottal source from a voiced sound is a difficult blind separation problem in speech signal processing. In this work, state-space methods are investigated to enhance the joint estimation of the glottal source and the vocal tract information. The aim of this paper is twofold. First, a stochastic glottal source is proposed, based on deterministic Liljencrants–Fant model and ruled by a stochastic difference equation. Such a representation allows to accurately capture any perturbation occurring at glottal level in real voices. A state-space voice model is formulated considering the stochastic glottal source. Then, combining this voice model and the state-space framework, an inverse filtering method is developed that allows to jointly estimate both glottal source and vocal tract filter. The performance of this method is studied by means of experiments with voices synthesized by applying both the source-filter theory and a physical based voice model. The method is also test using human voice signals. The results demonstrate that accurate estimates of the glottal source and the vocal tract filter can be obtained over several scenarios. Moreover, the method is shown to be robust with respect to different phonation types. Fil: Alzamendi, Gabriel Alejandro. Universidad Nacional de Entre Ríos. Facultad de Ingeniería. Departamento de Matemática e Informática. Laboratorio de Señales y Dinámicas no Lineales; Argentina. Consejo Nacional de Investigaciones Científicas y Técnicas. Centro de Investigaciones y Transferencia de Entre Ríos. Universidad Nacional de Entre Ríos. Centro de Investigaciones y Transferencia de Entre Ríos; Argentina Fil: Schlotthauer, Gaston. Universidad Nacional de Entre Ríos. Facultad de Ingeniería. Departamento de Matemática e Informática. Laboratorio de Señales y Dinámicas no Lineales; Argentina. Consejo Nacional de Investigaciones Científicas y Técnicas. Centro de Investigaciones y Transferencia de Entre Ríos. Universidad Nacional de Entre Ríos. Centro de Investigaciones y Transferencia de Entre Ríos; Argentina

This paper examines the applicability in realistic scenarios of two deep learning based solutions to the overlapping speaker separation problem. Firstly, we present experiments that show that these methods are applicable for a broad range of languages. Further experimentation indicates limited performance loss for untrained languages, when these have common features with the trained language(s). Secondly, it investigates how the methods deal with realistic background noise and proposes some modifications to better cope with these disturbances. The deep learning methods that will be examined are deep clustering and deep attractor networks., Comment: Interspeech 2019

Petri net synthesis deals with finding an unlabelled Petri net with a reachability graph isomorphic to a given usually finite labelled transition system (LTS). If there is no solution for a synthesis problem, we use label splitting. This means that we relabel edges until the LTS becomes synthesisable. We obtain an unlabelled Petri net and a relabelling function, which together form a labelled Petri net with the original, intended behaviour. By careful selection of the edges to relabel we hope to keep the alphabet of the LTS and the constructed Petri net as small as possible. Even approximation algorithms, not yielding an optimal relabelling, are hard to come by. Using region theory, we develop a polynomial heuristic based on two kinds of separation problems. These either demand distinct Petri net markings for distinct LTS states or a correspondence between the existence of an edge in the LTS and the activation of a transition under the state’s marking. If any separation problem is not solvable, relabelling of edges in the LTS becomes necessary. We show efficient ways to choose those edges.

International audience; We investigate a standard operator on classes of languages: unambiguous polynomial closure. We show that if C is a class of regular languages having some mild properties, the membership problem for its unambiguous polynomial closure UPol(C) reduces to the same problem for C. We give a new, self-contained and elementary proof of this result. We also show that unambiguous polynomial closure coincides with alternating left and right deterministic closure. Finally, if additionally C is finite, we show that the separation and covering problems are decidable for UPol(C).

We generalize the applicability of interactive methods for solving computationally demanding, that is, time-consuming, multiobjective optimization problems. For this purpose we propose a new agent assisted interactive algorithm. It employs a computationally inexpensive surrogate problem and four different agents that intelligently update the surrogate based on the preferences specified by a decision maker. In this way, we decrease the waiting times imposed on the decision maker during the interactive solution process and at the same time decrease the amount of preference information expected from the decision maker. The agent assisted algorithm is not specific to any interactive method or surrogate problem. As an example we implement our algorithm for the interactive NIMBUS method and the PAINT method for constructing the surrogate. This implementation was applied to support a real decision maker in solving a two-stage separation problem.

This paper introduces the anytime anywhere document analysis methodology applied in the context of computer-aided transcription. Its utility is revealed for documents which are difficult to analyse, as in the case of handwritten texts. A special focus lies on the glyph separation problem which turns out to be particularly complicated. As automatic methods show fundamental limitations, a number of interactive methods are proposed which are based on the interplay between user and machine. These methods get along without any assumptions concerning underlying languages or appearances of texts. An evaluation in the context of palaeography and applied to a well-established data set illustrates how well handwritings are dealt with, although they offer distinct differences in their regularity and shape.

In this paper, we consider a blind separation problem for single-channel mixtures of two QPSK signals, which have the same symbol rates. However, the present method is short of efficiency. To this question, we present a periodic linear multi-in multi-out model which is constructed by over-sampling. Under this condition, an algorithm based on lattice reduction (LR) and decision feedback equalization (DFE) is proposed for symbol detection. Simulation results are provided to illustrate the performance of the proposed algorithm.

The Vertex Separation Problem (VSP) is an NP-hard combinatorial optimization problem in the context of graph theory. The importance of studying VSP lies in its close relation with other problems. Thus, VSP has important practical applications in the contexts of very large scale integration design, computer language compiler design, natural language processing, order processing of manufactured products and bioinformatics. Up to our knowledge, there are only two trajectory-based metaheuristic algorithms for VSP documented in the literature. The main contribution of this chapter is that we extend the available heuristics to solve VSP by proposing a genetic algorithm (GA). It is of particular interest to study the impact of four different crossover operators in the algorithm performance. The experimental results showed that the order-based crossover is the best. Moreover, the best GA variant was compared with the best algorithm for VSP: GVNS. The results of this comparison showed that GVNS outperforms our best GA variant by approximately 1.54 times in solution quality.

In this paper, we address the problem of blind extraction and separation of a continuous chaotic signal from a linear mixture consisting of some chaotic signal and/or random signals. The problem of blind extraction is firstly formulated as a problem of the synchronization-based parameter estimation. Then an efficient least square based parameter estimation method is introduced to determine the desired extracting vector. The proposed blind signal extraction scheme is applicable to blind separation of chaotic signals by formulating the separation problem as the extraction of each chaotic source. Numerical simulation shows that the proposed approach can blindly extract and separate the desired chaotic signals and it is also robust to measurement noise.

A two-phase method for the pattern-driven recognition of objects in images is presented, implemented, and tested numerically. The method is based on the use of an active sensor. Possibilities for development are envisaged. This approach was shown to have advantages in solving the object-background separation problem and a high recognition rate was achieved with slow learning.

In this chapter, we consider finite transition systems exclusively. The chapter contains two sections. The first section shows that the net synthesis problem is NP-complete, hence one cannot construct very efficient synthesis algorithms. The second section constructs a flexible algorithm, based on minimal regions, that can be tailored to all forms of the net synthesis problem by selecting the relevant separation axioms. Most developments presented in the chapter, in spite of the title, are independent of elementary nets and valid in the larger framework of quasi-elementary nets. By nets and net systems, to restate the convention adopted in Chapter 1, we always mean quasi-elementary nets and quasi-elementary net systems, thus elementary is never meant unless it is explicit.

In order to separate a 3D chromatography, which is generated from High Performance Liquid Chromatography-Diode Array Detector (HPLC-DAD), into chromatograms and spectra, we proposed a model called parallel Independent Component Analysis constrained by Reference Curve (pICARC), which transforms the separation problem to a multi-parameter optimization issue. Then, A new algorithm named multi-areas Genetic Algorithm (mGA) is developed to search multiple solutions in parallel. It was demonstrated that our approach could successfully separate a given HPLC-DAD dataset into chromatograms and spectra with little errors and in short computational time.

We consider functionals which arise naturally from game theoretical consideration and are also generalizations of the functionals in Hisano and Maruyama (1989). In the study of the natural partial order on the set of these functionals, we establish a separation theorem, making use of the theorem of Hales-Jewett.

We provide efficient parallel algorithms for the minimum separation, offset range, and optimal offset problems for single-layer channel routing. We consider all the variations of these problems that are known to have linear-time sequential solutions rather than limiting attention to the "river-routing" context, where single-sided connections are disallowed. For the minimum separation problem, we obtain O( lg N) time on a CREW PRAM or [Formula: see text] time on a (common) CRCW PRAM, both with optimal work (processor-time product) of O(N), where $N$ is the number of terminals. For the offset range problem, we obtain the same time and processor bounds as long as only one side of the channel contains single-sided nets. For the optimal offset problem with single-sided nets on one side of the channel, we obtain time O( lg N lg lg N) on a CREW PRAM or [Formula: see text] time on a CRCW PRAM with O(N lg lg N) work. Not only does this improve on previous results for river routing, but we can obtain an even better time of O(( lg lg N)2) on the CRCW PRAM in the river routing context. In addition, wherever our results allow a channel boundary to contain single-sided nets, the results also apply when that boundary is ragged and N incorporates the number of bendpoints.

International audience; The Virtual Private Networks (VPN) optimal bandwidth allocation problem with tree toplogy has been widely discussed in the literature. Most of the algorithms proposed by researchers to solve this problem use approximation schemes. In this paper, we propose an exact and efficient Branch-and-Cut algorithm for the problem in the context of a hose workload model. In particular, we consider the case when the ingress and egress traffic at VPN endpoints are asymetric and the links of the network have unbounded capacities. The algorithm proposed here is based on a linear integer programming formulation for the problem introduced by Kumar et al. [2]. Using this and a deep investigation of the polyhedral structure of that formulation, our algorithm permits to solve large instances of the problem having up to 120 nodes and 10 terminals.

International audience; A separator for two languages is a third language containing the first one and disjoint from the second one. We investigate the following decision problem: given two regular input languages, decide whether there exists a locally testable (resp. a locally threshold testable) separator. In both cases, we design a decision procedure based on the occurrence of special patterns in automata accepting the input languages. We prove that the problem is computationally harder than deciding membership. The correctness proof of the algorithm yields a stronger result, namely a description of a possible separator. Finally, we discuss the same problem for context-free input~languages.

In this contribution, the optimization of periodic chromatographic simulated moving bed SMB processes is discussed. The rigorous optimization is based on a nonlinear pde model which incorporates rigorous models of the chromatographic columns and the discrete shifts of the inlet and outlet ports. The potential of the optimization is demonstrated for a separation problem with nonlinear isotherm of the Langmuir type for an SMB process and the ModiCon process. Here, an efficient numerical approach based on multiple shooting is employed. An overview of established optimization approaches for SMB processes is given.

In Chapter 1 Section 1.2.2, the mathematical expression of convolutive mixtures was introduced in the convolutive mixtures separation problem. In order to facilitate reading, basic expressions are cited again here.

In previous work, the constrained independent component analysis (cICA) algorithm has been proposed to extract the interested signals from the mixtures of some source signals. However, the simultaneous extraction of all signals at the same time presented by cICA prolongs the processing time of this algorithm to extract output signals. In this paper, we introduce a new version of the cICA algorithm to improve cICA in the computational time aspect. By whitening input signals, normalizing weight vectors, and using the one-by-one extraction of output signals, our proposed cICA algorithm has reduced the computational time to recover original signals when compared with the conventional cICA. Meanwhile our proposed cICA algorithm still retains the same recovering performance with that of the conventional cICA. Moreover, in this paper, we also introduce a potential application of our proposed cICA and the conventional cICA on the speech separation problem using priori information to extract the interested speech signals from mixed signals.

The acoustic echo cancellation (AEC) problem in the double-talk scenario and the blind source separation (BSS) problem resemble each other in that, for both problems, mixed signals are given and the objectives are to remove unwanted signals from the mixed signals. As many BSS algorithms have utilized the non-Gaussianity of the source signals to solve the separation problem, the super-Gaussianity of the near-end speech signal can be utilized to perform AEC in the double-talk scenario. Here, we propose a maximum likelihood (ML) approach using a super-Gaussian source prior to solve the double-talk-scenario AEC problem and compare the algorithm with minimizing mean squared error (MSE). The simulation results and analysis support the efficiency of the proposed method.

None of the alternative strategies for catalyst/product separation has yet reached the point where it can be commercialised for the rhodium catalysed hydroformyation of long chain alkenes and there are very few examples of commercialisation in any catalytic applications. Batch continuous processing with low pressure product distillation has been commercialised but the complexity of the system suggests that alternatives may be able to compete.

This paper introduces a Flexible ICA approach to a novel blind sources separation problem. The proposed on line algorithm performs the separation after the convolutive mixing of post nonlinear convolutive mixtures. The Flexibility of the algorithm is given by the on line estimation of the score function performed by Spline Neurons. Experimental results are described to show the effectiveness of the proposed technique.

Many neural networks for the blind separation problem have recently been proposed. However, in most of them, delays on input signals to networks are not considered. In this paper, we propose an extension of the network of Jutten and Herault (1991) that is applied to input signals including delays. Moreover, we present results of comparative simulations between the original Herault-Jutten network and our method for cases where the input signals are direct signals from sources and delay signals, such as reflections off walls.

This paper considers the separation of multiplexed chaotic and periodic pulse-trains. As the pulse-train source, the chaotic pulse-train generator is introduced and analyzed. In order to approach the separation problem, a master-slave network of the generators is studied and some of basic results are shown. Based on these results, a multiplex communication scheme is discussed.

The separation technologies of 3D chromatograms have been researched for a long time to obtain spectra and chromatogram peaks for individual compounds. However, before applying most of the current methods, the number of compounds must be known in advance. Independent Component Analysis (ICA) is applied to separate 3D chromatograms without knowing the compounds' number in advance, but the existence of the noise component in the results makes it complex for computation. In this paper, a parallel model of Independent Component Analysis constrained by a 5-parameter Reference Curve (pICA5pRC) is proposed based on the ICA model. Introducing a priori knowledge from chromatogram peaks, the pICA5pRC model transformed the 3D chromatogram separation problem to a 5 parameters optimization issue. An algorithm named multi-target particle swarm optimization (mPSO) has been developed to solve the pICA5pRC model. Through simulations, the performance and explanation of our method were described. Through experiments, the practicability of our method is validated. The results show that: (1) our method could separate 3D chromatograms efficiently even with severe overlap without knowing the compounds' number in advance; (2) our method extracted chromatogram peaks from the dataset directly without noise components; (3) our method could be applied to the practical HPLC-DAD dataset.

We present a methodology to estimate the maximum recovery of each component in an azeotropic multicomponent mixture processed at a batch distillation column. Based in the most recent advances in the field of separation feasibility we have adapted an algorithm to predict all the distillation regions for the multicomponent mixture with their corresponding natural and distillation boundaries. The results of the maximum separation problem togheter with those corresponding to conceptual dynamic runs are used to illustrate several important issues related to the synthesis and design of highly non-ideal azeotropic mixtures.

A numerical technique for simulating the dynamics and trajectory of multiple moving bodies is presented. A 6DOF model is fully integrated into the CFD solution procedure to determine the body dynamics. A highly automated Chimera technique has been developed and implemented into the software, CFD-FASTRAN, to achieve maximum flexibility and efficiency for CFD applications involving multiple moving bodies. The process is validated by application to a complex three dimensional store separation problem and compared against existing data. The code capability is further demonstrated by simulation of a missile tube launch.

Mixed model assembly lines are becoming increasingly used in manufacturing facilities to meet the diversified demands of customers. In this paper we deal with syncronous flow lines and wish to determine the job input sequence minimizing total tardiness. A ILP formulation of the problem is given. We prove the strong NP-hardness of the problem, we try to solve the linear relaxation of the problem and then introduce a number of cutting planes to solve the separation problem. Computational results confirm the effectiveness of the procedure.

The aim of this paper is to report on some initial studies which have been made into the use of hybrid analytical and experimental techniques in the stress separation problem for thermographic stress analysis using the SPATE system. The initial part of the study, which is reported here, is based upon the use of detailed finite element analysis of a number of selected SPATE benchmark problems to provide a comparison between the accuracy of the thermographic full field contouring and smoothing.

Impatience is frequently expressed at the slow rate of progress in improving the efficiency of air traffic control. It is being expressed most forcibly at this time in the United States, because the military services require the separation problem to be solved for the large numbers of military jet flights, and because civil jet aeroplanes have been ordered by the airlines and will soon appear on the Federal Airways. Although a great deal of basic study has been made of the problem of air traffic control, it is becoming clear that there is no panacea for its solution. Air traffic control is concerned with the safe expeditious flow of air traffic. This involves a number of separate but interrelated problems both in the aircraft and on the ground. Thus we see a programme of development on a very wide front involving navigation aids, communications, radar performance and presentation, data display and so on. Progress in any one field may affect the requirements in another, and so it is most necessary for all concerned to have a clear view of the situation as a whole in order that they may see the wood as well as the trees. Lively discussion of the subject can therefore be most fruitful, both among those directly concerned—the pilots and the controllers—and among the specialists in the various branches of development.

It is sometimes possible to obtain pure compounds after a single chromatographic step (e.g. preparative LC; Hostettmann et al. 1977 or RLCC ; Schubel and Stockigt 1984) but more often than not, a combination of methods is required. This fact becomes self-explanatory when one considers, for example, that the isolation of natural products from plant sources may involve the search for one compound out of an extract containing several thousand components. Of course, the particular separation problem depends on the discipline involved — obtention of a product from a synthetic reaction step may require the separation of only one by-product. As far as such relatively problem-free separations are concerned, however, other less expensive and less time-consuming techniques (crystallisation etc.) would be of first priority.

In column-chromatographic analysis of biological samples, complex chromatograms with long elution times are often encountered. Once the general separation conditions have been chosen, e.g. type and mode of use of chromatographic column and mode of detection, a number of continuously tunable variables can be used to optimize the overall analytical procedure. Necessarily a criterion has to be defined that permits an objective judgement of the resulting chromatograms. For this purpose chromatographic response functions (CRFs) used with formal strategies of experimental optimization have been developed. The strategies involve either entirely empirical search and mapping techniques or semi-empirical functions that account for specific molecular interactions. In any case the optimization procedure has to be chosen in accordance with the complexity of the separation problem, and the CRF must reflect fundamental analytical performance characteristics such as precision, accuracy and analysis time. Example in APPENDIX: thyronine I-enantiomers.

The problems of placement and routing in integrated circuit design have been gaining increasing attention as fabrication technology advances. Although a variety of those the problems have been proven to be.NP-hard, progress is being made on restricted versions. Tompa, for example, gives a quadratic solution to a particular single layer routing problem ([T]). This paper gives efficient algorithms for finding the separation and the offset in contexts which include his model.

There are basically three ways of classifying the more than one thousand scientific publications which describe the application of porous gels to the separation of substances. In one approach classes of substances are the guiding principle. In another attempt the experimental data are organized by areas of application. In Chapter 5 of this monograph a review will be presented which takes both these possibilities into consideration. Thirdly one could attempt — without consideration of classes of substances — to draw from practical experience in order to develop principles of application of gel chromatography and arrive at a classification. The latter approach has a didactic advantage. The reader does not have to familarize himself with an excessively large number of references and yet is in a position to obtain quickly information about possible approaches to the solution of a given separation problem. The present chapter is based on methodical aspects and is illustrated by a number of selected practical applications.