Cost-conscious comparison of supervised learning algorithms over multiple data sets

In the literature, there exist statistical tests to compare supervised learning algorithms on multiple data sets in terms of accuracy but they do not always generate an ordering. We propose Multi²Test, a generalization of our previous work, for ordering multiple learning algorithms on multiple data sets from “best” to “worst” where our goodness measure is composed of a prior cost term additional to generalization error. Our simulations show that Multi²Test generates orderings using pairwise tests on error and different types of cost using time and space complexity of the learning algorithms.     

FORECASTING OF MULTIVARIATE PERIODIC AUTOREGRESSIVE MOVING-AVERAGE PROCESSES

Abstract. Minimum mean square error forecasting of multivariate autoregressive moving-average processes with periodically varying parameters and orders is considered. General expressions are obtained for the forecasts, their errors and the covariance matrices of the forecast errors. Recursive evaluations of these quantities are shown to follow from the conditional expectation approach. Prediction ellipsoids and intervals for future values of the process are given. Update equations for the forecasts are obtained. The general results are illustrated and verified for a particular case of the process. A simulated example is given.     

Design and implementation of a personal computer based automaticvoltage regulator for a synchronous generator

An effective and straightforward design method for a digital voltage regulator for a synchronous generator is presented. The design is carried out directly in the z-domain using the root locus method. Simple software routines were written in assembly language thereby making it suitable for online computer control. Digital simulation studies were also carried out to test and compare the design and its dynamic results. The design was implemented on a laboratory size 5 kVA generator with an IBM personal computer acting as the digital controller. Results from the experimental set-ups reinforced the theoretical expectations     

Eigenclassifiers for combining correlated classifiers

In practice, classifiers in an ensemble are not independent. This paper is the continuation of our previous work on ensemble subset selection [A. Ula?, M. Semerci, O.T. Y?ld?z, E. Alpayd?n, Incremental construction of classifier and discriminant ensembles, Information Sciences, 179 (9) (2009) 1298–1318] and has two parts: first, we investigate the effect of four factors on correlation: (i) algorithms used for training, (ii) hyperparameters of the algorithms, (iii) resampled training sets, (iv) input feature subsets. Simulations using 14 classifiers on 38 data sets indicate that hyperparameters and overlapping training sets have higher effect on positive correlation than features and algorithms. Second, we propose postprocessing before fusing using principal component analysis (PCA) to form uncorrelated eigenclassifiers from a set of correlated experts. Combining the information from all classifiers may be better than subset selection where some base classifiers are pruned before combination, because using all allows redundancy.     

A novel method for reducing harmonics in series-connected rectifiers

A novel method for reducing harmonics in series-connected converters is proposed in this paper. Pulse multiplication for parallel-connected converters is usually easier to achieve because of the interphase transformer, which is absent in the series-connected converters. The proposed method in this paper adopts an averaging inductor to the series-connected converters and thus simplifies the pulse multiplication circuit greatly. The principle of this method is demonstrated by a three-thyristor switching circuit scheme, which can make a 12-pulse double bridge converter to operate at 36 pulses. Simulation results reinforce the theoretical work. Only slight changes are needed for the proposed method to reach higher pulse operations. This makes this method even more attractive when high-pulse operation is desired. The configurations of two-thyristor and four-thyristor switching schemes are also given.     

A search algorithm for the minimum cost covering of 0–1 integer sets

This article presents an algorithm for the realization of the minimum cover on a 0–1 integer set with linear constraints. This is done through a search vector which, after each step along the gradient of the objective function, realigns itself on the polyhedra formed by the constraint hypersurfaces. The technique yields convergence to the solution point through fast and straightforward computations. Computational results are given to demonstrate the concepts involved.     

Throughput Scalability of Wireless Hybrid Networks over a Random Geometric Graph

In this paper, we consider the transport capacity of ad hoc networks with a random flat topology under the present support of an infinite capacity infrastructure network. Such a network architecture allows ad hoc nodes to communicate with each other by purely using the remaining ad hoc nodes as their relays. In addition, ad hoc nodes can also utilize the existing infrastructure fully or partially by reaching any access point (or gateway) of the infrastructure network in a single or multi-hop fashion. Using the same tools as in [9], we show that the per source node capacity of Θ(W/log(N)) can be achieved in a random network scenario with the following assumptions: (i) The number of ad hoc nodes per access point is bounded above, (ii) each wireless node, including the access points, is able to transmit at W bits/sec using a fixed transmission range, and (iii) N ad hoc nodes, excluding the access points, constitute a connected topology graph. This is a significant improvement over the capacity of random ad hoc networks with no infrastructure support which is found as in [9]. We also show that even when less stringent requirements are imposed on topology connectivity, a per source node capacity figure that is arbitrarily close to Θ(1) cannot be obtained. Nevertheless, under these weak conditions, we can further improve per node throughput significantly. We also provide a limited extension on our results when the number of ad hoc nodes per access point is not bounded.Ulaş C. Kozat was born in 1975, in Adana, Turkey. He received his B.Sc. degree in Electrical and Electronics Engineering from Bilkent University, Ankara, Turkey and his M.Sc. in Electrical Engineering from The George Washington University, Washington D.C. in 1997 and 1999 respectively. He has received his Ph.D. degree in May 2004 from the Department of Electrical and Computer Engineering at University of Maryland, College Park. He has conducted research under the Institute for Systems Research (ISR) and Center for Hybrid and Satellite Networks (CSHCN) at the same university. He worked at HRL Laboratories and Telcordia Technologies Applied Research as a research intern. His current research interests primarily focus on wireless and hybrid networks that span multiple communication layers and networking technologies. Mathematical modelling, resource discovery and allocation, vertical integration of wireless systems and communication layers, performance analysis, architecture and protocol development are the main emphasis of his work. E-mail: kozat@isr.umd.eduLeandros Tassiulas (S′89, M′91) was born in 1965, in Katerini, Greece. He obtained the Diploma in Electrical Engineering from the Aristotelian University of Thessaloniki, Thessaloniki, Greece in 1987, and the M.S. and Ph.D. degrees in Electrical Engineering from the University of Maryland, College Park in 1989 and 1991 respectively.He is Professor in the Dept. of Computer and Telecommunications Engineering, University of Thessaly, Greece and Research Professor in the Dept. of Electrical and Computer Eng and the Institute for Systems Research, University of Maryland College Park since 2001. He has held positions as Assistant Professor at Polytechnic University New York (1991–95), Assistant and Associate Professor University of Maryland College Park (1995–2001) and Professor University of Ioannina Greece (1999–2001).His research interests are in the field of computer and communication networks with emphasis on fundamental mathematical models, architectures and protocols of wireless systems, sensor networks, high-speed internet and satellite communications.Dr. Tassiulas received a National Science Foundation (NSF) Research Initiation Award in 1992, an NSF CAREER Award in 1995 an Office of Naval Research, Young Investigator Award in 1997 and a Bodosaki Foundation award in 1999 and the INFOCOM′94 best paper award. E-mail: leandros@isr.umd.edu     

Hydrogels of polyvinylpyrrolidone (PVP) and poly(acrylic acid) (PAA) synthesized by photoinduced crosslinking of homopolymers

A novel method of covalent crosslinking between polyvinylpyrrolidone (PVP) and poly(acrylic acid) (PAA) resulting in hydrogels has been developed. The hydrogels were formed by photocrosslinking in oxygen-free aqueous solutions containing hydrogen peroxide as a source of hydroxyl radicals. The crosslinking was achieved via irradiation within the broad wavelength range from 200 to 800 nm, as well as by the light cut-off at λ > 300 nm. The obtained PAA-PVP gels were sensitive to pH. Protonation of the PAA carboxylic groups with decreasing pH promoted hydrogen bonding between the PAA and PVP segments within the crosslinked structure and caused the gel to collapse. This property enabled the use of the hydrogels as a simple chemical sensor. When loaded with glucose oxidase, the PAA-PVP gel's opacity and sedimentation due to the clearly observable phase separation were triggered by the presence of glucose due to a drop in pH caused by the enzymatic reaction.