Electrical Engineering

<p>Shiv Nadar University is a comprehensive, multidisciplinary, research-focused, and student-centric University offering a full range of academic programs at the undergraduate, postgraduate and doctoral level. With state-of-the-art infrastructure, the University comprises of academic wings, sophisticated labs, international standard sports facilities, amphitheaters, auditorium, conference rooms and smart classrooms. The University&rsquo;s goal is to become internationally recognized for the quality of its research and creative endeavors and their applicability to improving quality of life, generating new insights and expanding the boundaries of human knowledge creativity. Committed to excellence in teaching, research and service, the University aims to serve the higher education needs of India and the world beyond.</p>

Shiv Nadar University

This paper presents a linear SVM (Support Vector Machine) Pyramidal Tree (SVMPT) for binary classification tasks. SVMPT is a modified version of SVM based Tree Type Neural Networks (SVMTNN), reported earlier in the literature [1]. Both the algorithms use parameter-less SVM proposed by Mangasarian [2] for learning in each node. While SVMTNN insists on 100 percent training accuracy, linear SVMPT uses predetermined threshold value to determine when to stop adding new nodes. Experimental results on standard binary datasets show that the algorithm has good generalization capability, comparable to linear SVMs. We also present experimental results on extensions of linear SVMPT to multiclass datasets and datasets with fuzzy membership for each datapoint. © 2010 IEEE.

World University root node

(SoE) School of Engineering

DSDE 2010 - International Conference on Data Storage and Data Engineering

Binary classification using linear SVM pyramidal tree

We present an improved version of One-Against-All (OAA) method for multiclass SVM classification based on a decision tree approach. The proposed decision tree based OAA (DT-OAA) is aimed at increasing the classification speed of OAA by using posterior probability estimates of binary SVM outputs. DT-OAA decreases the average number of binary SVM tests required in testing phase to a greater extent when compared to OAA and other multiclass SVM methods. For a balanced multiclass dataset with K classes, under best situation, DT-OAA requires only (K + 1)/2 binary tests on an average as opposed to K binary tests in OAA; however, on imbalanced multiclass datasets we observed DT-OAA to be much faster with proper selection of order in which the binary SVMs are arranged in the decision tree. Computational comparisons on publicly available datasets indicate that the proposed method can achieve almost the same classification accuracy as that of OAA, but is much faster in decision making. © 2010 Springer Science+Business Media, LLC.

Neural Processing Letters

Fast multiclass SVM classification using decision tree based one-against-all method

Two-player zero-sum Markov game framework offers an effective platform for designing robust controllers. In the Markov game-based learning, theoretical convergence of the learning process with the function approximator cannot be guaranteed. However, fusing Q-learning with decision tree (DT) function approximator has shown good learning performance and more reliable convergence. It scales better to larger input spaces with lower memory requirements, and can solve problems that are infeasible using table lookup. This motivates us to introduce DT function approximator in Markov game reinforcement learning (RL) framework. This approach works, though it deals with only discrete actions. In realistic applications, it is imperative to deal with continuous state-action spaces. In this paper, we propose Markov game framework for continuous state-action space systems using fuzzy DT as a function approximator. Simulation experiments on a two-link robot manipulator bring out the importance of the proposed structure in terms of better robust performance and computational efficiency. Copyright © 2010 Inderscience Enterprises Ltd.

International Journal of Automation and Control

A fuzzy decision tree-based robust Markov game controller for robot manipulators

Thin films of Zn1-xCdxS (0.1 ≤ x ≤ 0.5) were prepared by using pulsed laser ablation technique on corning glass substrates. Phase transition from cubic to hexagonal in Zn1-xCdxS films is determined by X-ray diffraction analysis. We observed a lowering in the phase transition temperature with increase in the cadmium concentration. Transmission electron microscopy suggests the crystalline nature of thin films with average particle size of 15 nm. The grown Zn1-xCdxS samples show the high peak intensity ratio of the near band edge emission to the defect center luminescence even at room temperature, which indicates the small concentration of complex defects in the samples. Photoluminescence measurement show stoichiometric dependence of the energy band gap and is found to have quadratic dependence on x.

Journal of Nanoparticle Research

Photoluminescence measurements in the phase transition region of Zn 1-xCdxS films

"Strong Snapback" in DeNMOS transistors leads to weak ESD performance which is often represented by low It2 and strong die to die dependence. We report here the first experimental evidence that this can be controlled with introduction of source-resistance Rs. A new microscopic model has been analyzed to understand the physics of strong snapback and explain the experimental observations. Impact of current crowding phenomenon and role of adding a resistor across the source and ground has been broadly addressed in this paper. Also the current crowding phenomenon has been macroscopically modeled and a circuit model has been established. © 2010 IEEE.

IEEE International Reliability Physics Symposium Proceedings

Robust high current ESD performance of nano-meter scale DeNMOS by source ballasting

The paper presents a novel framework for learning the hash functions for indexing through Multiple Kernel Learning. The Distance Based Hashing function is applied which does the object projection to hash space by preserving inter object distances. In recent works, the kernel matrix has been proved to be more accurate representation of similarity in various recognition problems. Our framework learns the optimal kernel for hashing by parametrized linear combination of base kernels. A novel application of Genetic Algorithm for the optimization of kernel combination parameters is presented. We also define new texture based feature representation for images. Our proposed framework can also be applied for optimal combination of multiple sources for indexing. The evaluation of the proposed framework is presented for CIFAR-10 dataset 1 by applying individual and combination of different features. Additionally, the primary experimental results with MNIST dataset2 is also presented. © 2010 ACM.

ACM International Conference Proceeding Series

Multiple kernel learning for image indexing

Zn 0.9Cd 0.1S nanoparticles doped with 0.005-0.24 M cobalt have been prepared by co-precipitation technique in ice bath at 280 K. For the cobalt concentration &gt;0.18 M, XRD pattern shows unidentified phases along with Zn 0.9Cd 0.1S sphalerite phase. For low cobalt concentration (≤0.05 M) particle size, d XRD is ~3.5 nm, while for high cobalt concentration (&gt;0.05 M) particle size decreases abruptly (~2 nm) as detected by XRD. However, TEM analysis shows the similar particle size (~3.5 nm) irrespective of the cobalt concentration. Local strain in the alloyed nanoparticles with cobalt concentration of 0.18 M increases ~46% in comparison to that of 0.05 M. Direct to indirect energy band-gap transition is obtained when cobalt concentration goes beyond 0.05 M. A red shift in energy band gap is also observed for both the cases. Nanoparticles with low cobalt concentrations were found to have paramagnetic nature with no antiferromagnetic coupling. A negative Curie-Weiss temperature of -75 K with antiferromagnetic coupling was obtained for the high cobalt concentration. © 2009 to the authors.

Fulltext

Nanoscale Research Letters

Influence of cobalt doping on the physical properties of Zn 0.9Cd 0.1S Nanoparticles

We propose knowledge based versions of a relatively new family of SVM algorithms based on two non-parallel hyperplanes. Specifically, we consider prior knowledge in the form of multiple polyhedral sets and incorporate the same into the formulation of linear Twin SVM (TWSVM)/Least Squares Twin SVM (LSTWSVM) and term them as knowledge based TWSVM (KBTWSVM)/knowledge based LSTWSVM (KBLSTWSVM). Both of these formulations are capable of generating non-parallel hyperplanes based on real-world data and prior knowledge. We derive the solution of KBLSTWSVM and use it in our computational experiments for comparison against other linear knowledge based SVM formulations. Our experiments show that KBLSTWSVM is a versatile classifier whose solution is extremely simple when compared with other linear knowledge based SVM algorithms. © 2010 Elsevier Inc. All rights reserved.

Information Sciences

Knowledge based Least Squares Twin support vector machines

Reinforcement learning (RL) has now evolved as a major technique for adaptive optimal control of nonlinear systems. However, majority of the RL algorithms proposed so far impose a strong constraint on the structure of environment dynamics by assuming that it operates as a Markov decision process (MDP). An MDP framework envisages a single agent operating in a stationary environment thereby limiting the scope of application of RL to control problems. Recently, a new direction of research has focused on proposing Markov games as an alternative system model to enhance the generality and robustness of the RL based approaches. This paper aims to present this new direction that seeks to synergize broad areas of RL and Game theory, as an interesting and challenging avenue for designing intelligent and reliable controllers. First, we briefly review some representative RL algorithms for the sake of completeness and then describe the recent direction that seeks to integrate RL and game theory. Finally, open issues are identified and future research directions outlined. © 2009 Elsevier B.V. All rights reserved.

Applied Soft Computing Journal

Synergizing reinforcement learning and game theory-A new direction for control

Dynamic axial displacement of high voltage (HV) winding disk of a power transformer has been calculated by Green's function approach, when time varying axial electromagnetic forces corresponding to worst case of short circuit condition, are applied. An elaborate analysis of the axial electromagnetic forces and corresponding axial displacement of a top-end disk of HV winding is carried out. Under the action of these time varying forces, the winding moves in the axial direction as a function of time between radially placed insulation spacers. Between the spacer supports the winding sections can be considered as clamped-clamped (C-C) beams. Static Green's function of such a beam structure is determined first, and convolving it with axial electromagnetic force determines the axial displacement (bending) at any point on the beam for the purpose of verification of the methodology. Dynamic Green's function of the same winding sector is subsequently derived, which is used to obtain displacement of the winding section by convolving it with an applied time varying axial electromagnetic force. The results are verified with commercial FEM software. © 2010 - IOS Press and the authors.