Efficient algorithms for learning kernels from multiple similarity matrices with general convex loss functions

In this paper we consider the problem of learning an n × n kernel matrix from m(1) similarity matrices under general convex loss. Past research have extensively studied the m = 1 case and have derived several algorithms which require sophisticated techniques like ACCP, SOCP, etc. The existing algorithms do not apply if one uses arbitrary losses and often can not handle m > 1 case. We present several provably convergent iterative algorithms, where each iteration requires either an SVM or a Multiple Kernel Learning (MKL) solver for m > 1 case. One of the major contributions of the paper is to extend the well knownMirror Descent(MD) framework to handle Cartesian product of psd matrices. This novel extension leads to an algorithm, called EMKL, which solves the problem in O(m2 log n 2) iterations; in each iteration one solves an MKL involving m kernels and m eigen-decomposition of n × n matrices. By suitably defining a restriction on the objective function, a faster version of EMKL is proposed, called REKL,which avoids the eigen-decomposition. An alternative to both EMKL and REKL is also suggested which requires only an SVMsolver. Experimental results on real world protein data set involving several similarity matrices illustrate the efficacy of the proposed algorithms.

Space vector based bus clamped PWM algorithms for three level inverters: implementation, performance analysis and application considerations

This paper describes the different types of space vector based bus clamped PWM algorithms for three level inverters. A novel bus clamp PWM algorithm for low modulation indices region is also presented. The principles and switching sequences of all the types of bus clamped algorithms for high switching frequency are presented. Synchronized version of the PWM sequences for high power applications where switching frequency is low is also presented. The implementation details on DSP based digital controller and experimental results are presented. The THD of the output waveforms is studied for the entire operating region and is compared with the conventional space vector PWM technique. The bus clamped techniques can be used to reduce the switching losses or to improve the output voltage quality or both.. Different issues dominate depending on the type of application and power rating of the inverters. The results presented in this paper can be used for judicious use of the PWM techniques, which result in improved system efficiency and performance.

Practical fully three-dimensional reconstruction algorithms for diffuse optical tomography

We have developed an efficient fully three-dimensional (3D) reconstruction algorithm for diffuse optical tomography (DOT). The 3D DOT, a severely ill-posed problem, is tackled through a pseudodynamic (PD) approach wherein an ordinary differential equation representing the evolution of the solution on pseudotime is integrated that bypasses an explicit inversion of the associated, ill-conditioned system matrix. One of the most computationally expensive parts of the iterative DOT algorithm, the reevaluation of the Jacobian in each of the iterations, is avoided by using the adjoint-Broyden update formula to provide low rank updates to the Jacobian. In addition, wherever feasible, we have also made the algorithm efficient by integrating along the quadratic path provided by the perturbation equation containing the Hessian. These algorithms are then proven by reconstruction, using simulated and experimental data and verifying the PD results with those from the popular Gauss-Newton scheme. The major findings of this work are as follows: (i) the PD reconstructions are comparatively artifact free, providing superior absorption coefficient maps in terms of quantitative accuracy and contrast recovery; (ii) the scaling of computation time with the dimension of the measurement set is much less steep with the Jacobian update formula in place than without it; and (iii) an increase in the data dimension, even though it renders the reconstruction problem less ill conditioned and thus provides relatively artifact-free reconstructions, does not necessarily provide better contrast property recovery. For the latter, one should also take care to uniformly distribute the measurement points, avoiding regions close to the source so that the relative strength of the derivatives for measurements away from the source does not become insignificant. (c) 2012 Optical Society of America

Fusion of algorithms for compressed sensing

For compressed sensing (CS), we develop a new scheme inspired by data fusion principles. In the proposed fusion based scheme, several CS reconstruction algorithms participate and they are executed in parallel, independently. The final estimate of the underlying sparse signal is derived by fusing the estimates obtained from the participating algorithms. We theoretically analyze this fusion based scheme and derive sufficient conditions for achieving a better reconstruction performance than any participating algorithm. Through simulations, we show that the proposed scheme has two specific advantages: 1) it provides good performance in a low dimensional measurement regime, and 2) it can deal with different statistical natures of the underlying sparse signals. The experimental results on real ECG signals shows that the proposed scheme demands fewer CS measurements for an approximate sparse signal reconstruction.

Diffraction tomography from intensity measurements: an evolutionary stochastic search to invert experimental data

We develop iterative diffraction tomography algorithms, which are similar to the distorted Born algorithms, for inverting scattered intensity data. Within the Born approximation, the unknown scattered field is expressed as a multiplicative perturbation to the incident field. With this, the forward equation becomes stable, which helps us compute nearly oscillation-free solutions that have immediate bearing on the accuracy of the Jacobian computed for use in a deterministic Gauss-Newton (GN) reconstruction. However, since the data are inherently noisy and the sensitivity of measurement to refractive index away from the detectors is poor, we report a derivative-free evolutionary stochastic scheme, providing strictly additive updates in order to bridge the measurement-prediction misfit, to arrive at the refractive index distribution from intensity transport data. The superiority of the stochastic algorithm over the GN scheme for similar settings is demonstrated by the reconstruction of the refractive index profile from simulated and experimentally acquired intensity data. (C) 2014 Optical Society of America

Efficient simulation of unitary operators by combining two numerical algorithms: An NMR simulation of the mirror-inversion propagator of an XY spin chain

Precise experimental implementation of unitary operators is one of the most important tasks for quantum information processing. Numerical optimization techniques are widely used to find optimized control fields to realize a desired unitary operator. However, finding high-fidelity control pulses to realize an arbitrary unitary operator in larger spin systems is still a difficult task. In this work, we demonstrate that a combination of the GRAPE algorithm, which is a numerical pulse optimization technique, and a unitary operator decomposition algorithm Ajoy et al., Phys. Rev. A 85, 030303 (2012)] can realize unitary operators with high experimental fidelity. This is illustrated by simulating the mirror-inversion propagator of an XY spin chain in a five-spin dipolar coupled nuclear spin system. Further, this simulation has been used to demonstrate the transfer of entangled states from one end of the spin chain to the other end.

New signal processing algorithms for automated external defibrillators

[ES]La fibrilación ventricular (VF) es el primer ritmo registrado en el 40\,\% de las muertes súbitas por paro cardiorrespiratorio extrahospitalario (PCRE). El único tratamiento eficaz para la FV es la desfibrilación mediante una descarga eléctrica. Fuera del hospital, la descarga se administra mediante un desfibrilador externo automático (DEA), que previamente analiza el electrocardiograma (ECG) del paciente y comprueba si presenta un ritmo desfibrilable. La supervivencia en un caso de PCRE depende fundamentalmente de dos factores: la desfibrilación temprana y la resucitación cardiopulmonar (RCP) temprana, que prolonga la FV y por lo tanto la oportunidad de desfibrilación. Para un correcto análisis del ritmo cardiaco es necesario interrumpir la RCP, ya que, debido a las compresiones torácicas, la RCP introduce artefactos en el ECG. Desafortunadamente, la interrupción de la RCP afecta negativamente al éxito en la desfibrilación. En 2003 se aprobó el uso del DEA en pacientes entre 1 y 8 años. Los DEA, que originalmente se diseñaron para pacientes adultos, deben discriminar de forma precisa las arritmias pediátricas para que su uso en niños sea seguro. Varios DEAs se han adaptado para uso pediátrico, bien demostrando la precisión de los algoritmos para adultos con arritmias pediátricas, o bien mediante algoritmos específicos para arritmias pediátricas. Esta tesis presenta un nuevo algoritmo DEA diseñado conjuntamente para pacientes adultos y pediátricos. El algoritmo se ha probado exhaustivamente en bases de datos acordes a los requisitos de la American Heart Association (AHA), y en registros de resucitación con y sin artefacto RCP. El trabajo comenzó con una larga fase experimental en la que se recopilaron y clasificaron retrospectivamente un total de 1090 ritmos pediátricos. Además, se revisó una base de arritmias de adultos y se añadieron 928 nuevos ritmos de adultos. La base de datos final contiene 2782 registros, 1270 se usaron para diseñar el algoritmo y 1512 para validarlo. A continuación, se diseñó un nuevo algoritmo DEA compuesto de cuatro subalgoritmos. Estos subalgoritmos están basados en un conjunto de nuevos parámetros para la detección de arritmias, calculados en diversos dominios de la señal, como el tiempo, la frecuencia, la pendiente o la función de autocorrelación. El algoritmo cumple las exigencias de la AHA para la detección de ritmos desfibrilables y no-desfibrilables tanto en pacientes adultos como en pediátricos. El trabajo concluyó con el análisis del comportamiento del algoritmo con episodios reales de resucitación. En los ritmos que no contenían artefacto RCP se cumplieron las exigencias de la AHA. Posteriormente, se estudió la precisión del algoritmo durante las compresiones torácicas, antes y después de filtrar el artefacto RCP. Para suprimir el artefacto se utilizó un nuevo método desarrollado a lo largo de la tesis. Los ritmos desfibrilables se detectaron de forma precisa tras el filtrado, los no-desfibrilables sin embargo no.

Robot navigation in dense crowds : statistical models and experimental studies of human robot cooperation

This thesis explores the problem of mobile robot navigation in dense human crowds. We begin by considering a fundamental impediment to classical motion planning algorithms called the freezing robot problem: once the environment surpasses a certain level of complexity, the planner decides that all forward paths are unsafe, and the robot freezes in place (or performs unnecessary maneuvers) to avoid collisions. Since a feasible path typically exists, this behavior is suboptimal. Existing approaches have focused on reducing predictive uncertainty by employing higher fidelity individual dynamics models or heuristically limiting the individual predictive covariance to prevent overcautious navigation. We demonstrate that both the individual prediction and the individual predictive uncertainty have little to do with this undesirable navigation behavior. Additionally, we provide evidence that dynamic agents are able to navigate in dense crowds by engaging in joint collision avoidance, cooperatively making room to create feasible trajectories. We accordingly develop interacting Gaussian processes, a prediction density that captures cooperative collision avoidance, and a "multiple goal" extension that models the goal driven nature of human decision making. Navigation naturally emerges as a statistic of this distribution.

Most importantly, we empirically validate our models in the Chandler dining hall at Caltech during peak hours, and in the process, carry out the first extensive quantitative study of robot navigation in dense human crowds (collecting data on 488 runs). The multiple goal interacting Gaussian processes algorithm performs comparably with human teleoperators in crowd densities nearing 1 person/m², while a state of the art noncooperative planner exhibits unsafe behavior more than 3 times as often as the multiple goal extension, and twice as often as the basic interacting Gaussian process approach. Furthermore, a reactive planner based on the widely used dynamic window approach proves insufficient for crowd densities above 0.55 people/m². We also show that our noncooperative planner or our reactive planner capture the salient characteristics of nearly any dynamic navigation algorithm. For inclusive validation purposes, we show that either our non-interacting planner or our reactive planner captures the salient characteristics of nearly any existing dynamic navigation algorithm. Based on these experimental results and theoretical observations, we conclude that a cooperation model is critical for safe and efficient robot navigation in dense human crowds.

Finally, we produce a large database of ground truth pedestrian crowd data. We make this ground truth database publicly available for further scientific study of crowd prediction models, learning from demonstration algorithms, and human robot interaction models in general.

Optimization of positions and currents of tokamak poloidal field coils using genetic algorithms

Plasma equilibrium geometry has a great influence on the confinement and magnetohydrodynamic stability in tokamaks. The poloidal field (PF) system of a tokamak should be optimized to support the prescribed plasma equilibrium geometry. In this paper, a genetic algorithm-based method is applied to solve the optimization of the positions and currents of tokamak PF coils. To achieve this goal, we first describe the free-boundary code EQT Based on the EQT code, a genetic algorithm-based method is introduced to the optimization. We apply this new method to the PF system design of the fusion-driven subcritical system and plasma equilibrium geometry optimization of the Experimental Advanced Superconducting Tokamak (EAST). The results indicate that the optimization of the plasma equilibrium geometry can be improved by using this method.

Crystals that count! Physical principles and experimental investigations of DNA tile self-assembly

Algorithmic DNA tiles systems are fascinating. From a theoretical perspective, they can result in simple systems that assemble themselves into beautiful, complex structures through fundamental interactions and logical rules. As an experimental technique, they provide a promising method for programmably assembling complex, precise crystals that can grow to considerable size while retaining nanoscale resolution. In the journey from theoretical abstractions to experimental demonstrations, however, lie numerous challenges and complications.

In this thesis, to examine these challenges, we consider the physical principles behind DNA tile self-assembly. We survey recent progress in experimental algorithmic self-assembly, and explain the simple physical models behind this progress. Using direct observation of individual tile attachments and detachments with an atomic force microscope, we test some of the fundamental assumptions of the widely-used kinetic Tile Assembly Model, obtaining results that fit the model to within error. We then depart from the simplest form of that model, examining the effects of DNA sticky end sequence energetics on tile system behavior. We develop theoretical models, sequence assignment algorithms, and a software package, StickyDesign, for sticky end sequence design.

As a demonstration of a specific tile system, we design a binary counting ribbon that can accurately count from a programmable starting value and stop growing after overflowing, resulting in a single system that can construct ribbons of precise and programmable length. In the process of designing the system, we explain numerous considerations that provide insight into more general tile system design, particularly with regards to tile concentrations, facet nucleation, the construction of finite assemblies, and design beyond the abstract Tile Assembly Model.

Finally, we present our crystals that count: experimental results with our binary counting system that represent a significant improvement in the accuracy of experimental algorithmic self-assembly, including crystals that count perfectly with 5 bits from 0 to 31. We show some preliminary experimental results on the construction of our capping system to stop growth after counters overflow, and offer some speculation on potential future directions of the field.

Phase unwrapping algorithms, respectively, based on path-following and discrete cosine transform

This paper describes a path-following phase unwrapping algorithm and a phase unwrapping algorithm based on discrete cosine transform (DCT) which accelerates the Computation and suppresses the propagation of noise. Through analysis of fringe pattern with serious noises simulated in mathematic model, we make a contrast between path-following algorithm and DCT algorithm. The advantages and disadvantages or analytical fringe pattern are also given through comparison of two algorithms. Three-dimensional experimental results have been given to prove the validity of these algorithms. Despite DCT phase unwrapping technique robustness and speed in some cases, it cannot be unwrapping inconsistencies phase. The path-following algorithm can be used in automation analysis of fringe patterns with little influence of noise. (c) 2007 Elsevier GmbH. All rights reserved.

Distributed flow algorithms for scalable similarity visualization

We describe simple yet scalable and distributed algorithms for solving the maximum flow problem and its minimum cost flow variant, motivated by problems of interest in objects similarity visualization. We formulate the fundamental problem as a convex-concave saddle point problem. We then show that this problem can be efficiently solved by a first order method or by exploiting faster quasi-Newton steps. Our proposed approach costs at most O(|ε|) per iteration for a graph with |ε| edges. Further, the number of required iterations can be shown to be independent of number of edges for the first order approximation method. We present experimental results in two applications: mosaic generation and color similarity based image layouting. © 2010 IEEE.

New algorithms of the TSM and TOM methods for calibrating microwave test fixtures

Based on the conventional through-short-match (TSM) method, an improved TSM method has been proposed in this Letter. This method gives an analytical solution and has almost all the advantages of conventional TSM methods. For example, it has no phase uncertainty and no bandwidth limitation. The experimental results show that the accuracy can be significantly improved with this method. The proposed theory can be applied to the through-open-match (TOM) method. (C) 2002 Wiley Periodicals. Inc.

Algorithms for indoor localization based on IEEE 802.15.4-2011 UWB and inertial sensors

In this thesis, extensive experiments are firstly conducted to characterize the performance of using the emerging IEEE 802.15.4-2011 ultra wideband (UWB) for indoor localization, and the results demonstrate the accuracy and precision of using time of arrival measurements for ranging applications. A multipath propagation controlling technique is synthesized which considers the relationship between transmit power, transmission range and signal-to-noise ratio. The methodology includes a novel bilateral transmitter output power control algorithm which is demonstrated to be able to stabilize the multipath channel, and enable sub 5cm instant ranging accuracy in line of sight conditions. A fully-coupled architecture is proposed for the localization system using a combination of IEEE 802.15.4-2011 UWB and inertial sensors. This architecture not only implements the position estimation of the object by fusing the UWB and inertial measurements, but enables the nodes in the localization network to mutually share positional and other useful information via the UWB channel. The hybrid system has been demonstrated to be capable of simultaneous local-positioning and remote-tracking of the mobile object. Three fusion algorithms for relative position estimation are proposed, including internal navigation system (INS), INS with UWB ranging correction, and orientation plus ranging. Experimental results show that the INS with UWB correction algorithm achieves an average position accuracy of 0.1883m, and gets 83% and 62% improvements on the accuracy of the INS (1.0994m) and the existing extended Kalman filter tracking algorithm (0.5m), respectively.

Iterative inversion of experimental data in weighted spaces

We present iterative algorithms for solving linear inverse problems with discrete data and compare their performances with the method of singular function expansion, in view of applications in optical imaging and particle sizing.