Traditional error-correcting output codes (ECOCs) decompose a multi-class classification problem into many binary problems. Although it seems natural to use ECOCs for multi-label problems as well, doing so naively cre...
详细信息
Traditional error-correcting output codes (ECOCs) decompose a multi-class classification problem into many binary problems. Although it seems natural to use ECOCs for multi-label problems as well, doing so naively creates issues related to: the validity of the encoding, the efficiency of the decoding, the predictability of the generated codeword, and the exploitation of the label dependency. Using canonical correlation analysis, we propose an error-correcting code for multi-label classification. Label dependency is characterized as the most predictable directions in the label space, which are extracted as canonical output variates and encoded into the codeword. Predictions for the codeword define a graphical model of labels with both Bernoulli potentials (from classifiers on the labels) and Gaussian potentials (from regression on the canonical output variates). Decoding is performed by mean-field approximation. We establish connections between the proposed code and research areas such as compressed sensing and ensemble learning. Some of these connections contribute to better understanding of the new code, and others lead to practical improvements in code design. In our empirical study, the proposed code leads to substantial improvements compared to various competitors in music emotion classification and outdoor scene recognition. Copyright 2011 by the authors.
Non-Negative Matrix Factorization (NMF) is a dimensionality reduction method that has been shown to be very useful for a variety of tasks in machine learning and data mining. One of the fastest algorithms for NMF is t...
详细信息
Non-Negative Matrix Factorization (NMF) is a dimensionality reduction method that has been shown to be very useful for a variety of tasks in machine learning and data mining. One of the fastest algorithms for NMF is the Block Principal Pivoting method (BPP) of (Kim & Park, 2008b), which follows a block coordinate descent approach. the optimization in each iteration involves solving a large number of expensive least squares problems. Taking the view that the design matrix was generated by a stochastic process, and using the asymptotic normality of the least squares estimator, we propose a method for improving the performance of the BPP method. Our method starts with a small subset of the columns and rows of the original matrix and uses frequentist hypothesis tests to adaptively increase the size of the problem. this achieves two objectives: 1) during the initial phase of the algorithm we solve far fewer, much smaller sized least squares problems and 2) all hypothesis tests failing while using all the data represents a principled, automatic stopping criterion. Experiments on three real world datasets show that our algorithm significantly improves the performance of the original BPP algorithm. Copyright 2011 by the authors.
Reduction of structural complexity of digital systems has been emerging as one of the major areas of concern to the system designers. the deployment of any physical circuit or system in any specific application is sol...
详细信息
Reduction of structural complexity of digital systems has been emerging as one of the major areas of concern to the system designers. the deployment of any physical circuit or system in any specific application is solely determined by the intricacy of the concerned circuit which restricts the use of sophisticated circuits in high speed operation. thus the design of computationally efficient structure has drawn special attention to the modern researchers. Due to the immense development of evolutionary computation techniques over the last few years, they are being increasingly used for a number of signal processing applications. this paper deals with one such robust technique called Differential Evolution (DE) and the impact of its different mutation strategies for efficient design of multiplier-less low-pass Finite duration Impulse Response (FIR) filter. In order to find out the most favourable mutation scheme, the computational efficiency of various mutation schemes has been studied. For this purpose, the convergence behaviour and error histogram of DE algorithm has been presented. From the simulation results, the name of the computationally most efficient mutation strategy has been suggested. Finally, the superiority of the established mutation strategy has been reconfirmed in terms of the filter performance.
Heart disease is the leading cause of death in both men and women claiming up to 900,000 lives every year. Worldwide. An irregular heartbeat is an arrhythmia (also called dysrhythmia).Up to half of these deaths occurs...
详细信息
Heart disease is the leading cause of death in both men and women claiming up to 900,000 lives every year. Worldwide. An irregular heartbeat is an arrhythmia (also called dysrhythmia).Up to half of these deaths occurs even before emergency services can step in to intervene. To achieve this goal in this paper we have detected the peak of the QRS signal by using QRS detection method of Murthy and Rangaraj which is derivative based method, describe in Verilog HDL. the generated source has been simulated for validation and tested on software Verilogger Pro6.5 demo *** best of the authers' knowledge this is the first attempt propose design a digital system of the QRS detection used by Murthy and Rangaraj algorithm which can be further implemented by the manufacturer company. In test bench coding we have given some sample values as an input of the peak detecting equations of Murthy and Rangaraj method and have got the appropriate output values according to the equations. From the output values we got the key point of RR interval between the peaks of QRS signal.
this paper deals with a three-phase inverter controller design for monobloc configuration of small swirl turbine. the aim is an phasing of the system to grid. A simulation design is described which leads to cascade co...
详细信息
this paper deals with a three-phase inverter controller design for monobloc configuration of small swirl turbine. the aim is an phasing of the system to grid. A simulation design is described which leads to cascade controller with simple structure: lower loop is used for amplitude control and upper loop for phase control. First the highly nonlinear detailed three-phase inverter with switching components and diodes is modeled. then based on RMS measurement a compensated LTI model is extracted which is used by H-infinity loop-shaping controller synthesis for both controllers design. Lastly design results are verified by simulations applying original three-phase inverter model taking into account high level of disturbances. Resulting system meets with reserve desired behaviour.
In this paper, a complete novel and unique concept of Adaptive Cyclic Prefix (ACP) is proposed for IEEE 802.16e-WiMAX Physical Layer (PHY) using a Simulink-VSA based simulation model. Implementation of ACP algorithm p...
详细信息
In this paper, a complete novel and unique concept of Adaptive Cyclic Prefix (ACP) is proposed for IEEE 802.16e-WiMAX Physical Layer (PHY) using a Simulink-VSA based simulation model. Implementation of ACP algorithm provides a better QoS in the form of lowest average error for low channel SNR condition, compared to the existing Fixed Cyclic Prefix (FCP) in WiMAX. MATLAB plays the crucial part for implementing ACP, while the WiMAX PHY Layer scenario is created within Simulink workspace. By measuring Error Vector Magnitude (EVM) and Relative Constellation Error (RCE) in Vector Signal Analyser (VSA), a significant improvement is obtained in average error performance with respect to varying Channel SNR for ACP compared to FCP. Simulation results also show that ACP is efficient enough to be used for each of the modulation and coding specified for a WiMAX system, making it an easy alternative to the standard Adaptive Modulation and Coding (AMC). Finally exhaustive simulation results have been included in support to the success of the algorithm.
Recently, methods that combine the concepts of Dynamic Voltage Scaling (DVS) and Dynamic Power Management (DPM) approaches have been proposed for a single task. this paper proposes Adaptive Power Management System (AP...
详细信息
Recently, methods that combine the concepts of Dynamic Voltage Scaling (DVS) and Dynamic Power Management (DPM) approaches have been proposed for a single task. this paper proposes Adaptive Power Management System (APMS), which utilizes useful features of both DPM and DVS and an existing pattern analysis algorithm, and new BET-based Task Partition Scheduling (BTPS). In order to determine the optimal processor speed that minimizes power consumption, the extra consumption required by the scheduler is compared withthat reduced by BTPS. the proposed approach demonstrates better performance compared to existing power management schemes for not only a single task but multiple tasks.
Most of the energy consumed by the heating industry is from the burning of non-renewable coal. the environment problems caused by burning coal are more and more serious. the development of WSN and the advent of micro-...
详细信息
Most of the energy consumed by the heating industry is from the burning of non-renewable coal. the environment problems caused by burning coal are more and more serious. the development of WSN and the advent of micro-controller chip integrated with RF module make it possible to design a low-cost, easy-deployed and automatic user terminal for central heating system. this paper presents the design and implementation of this kind of automatic monitoring and control system. It provides the ability of not only thermal measurement but also automatic control. Using traditional central heating system, known as district heating system, users can not control their room temperature; but using this system they can. this will explore the users' potential for saving energy.
In this paper design of nonlinear sliding mode feedback controller for a model representing crowd dynamics is presented. the model is presented as hyperbolic partial differential equation based on the laws of conserva...
详细信息
In this paper design of nonlinear sliding mode feedback controller for a model representing crowd dynamics is presented. the model is presented as hyperbolic partial differential equation based on the laws of conservation of mass. the feedback control is designed in presence of uncertainties due to parametric variations. the controller is designed using sliding mode method. the controllers designed are shown to be robust to parametric variation.
暂无评论