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.     

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.     

A stochastic neighborhood search approach for airport gate assignment problem

An appropriate and efficient gate assignment is of great importance in airports since it plays a major role in the revenue obtained from the airport operations. In this study, we have focused mainly on maximum gate employment, or in other words minimize the total duration of un-gated flights. Here, we propose a method that combines the benefits of heuristic approaches with some stochastic approach instead of using a purely probabilistic approach to top-down solution of the problem. The heuristic approaches are usually used in order to provide a fast solution of the problem and later stochastic searches are used in order to ameliorate the previous results of the heuristic approach whenever possible. The proposed method generates an assignment order for the whole planes that corresponds to assignment priority. The ordering process is followed by the allocation step. Since, in practice, each airport has its own physical architecture, there have been arisen many constraints mainly concerning airplane types and parking lots in this step. Sequentially handling the plane ordering and allocation phases provides us great modularity in handling the constraints. The effectiveness of the proposed methodology has been tried to be illustrated firstly on fictively generated flight schedule data and secondly on the real world data obtained from a real world application developed for ?stanbul Atatürk Airport.     

The effect of heat treatment on high temperature mechanical properties of microalloyed medium carbon steel

In the present work, high temperature tensile properties and abrasive wear performance of a microalloyed medium carbon steel has been examined. Tensile and abrasive wear testing were carried out on as-received and heat treated specimens. The research has shown that microalloyed medium carbon steel was susceptible to dynamic strain ageing due to interaction of mobile dislocations and solid atoms, such as carbon and/or nitrogen. The interaction between dislocations and solid atoms at 200–300 °C changes the work hardening rate and contributes to dynamic strain ageing. These interactions also increased abrasive wear resistance of the microalloyed medium carbon steel at 300 °C. Therefore, the inference can be drawn that dynamic strain ageing caused an improvement on abrasion resistance.     

Calculations of the Elastic Scattering Properties of Ultra-cold Francium Isotopes

The calculations of the elastic scattering properties of ultra-cold francium isotopes are reported in detail. A parametric model potential for the triplet molecular states of the Fr₂ is represented, and the scattering lengths a_t and the effective range r_e are calculated using WKB and Numerov methods for Fr--Fr in the triplet state. The convergence of these scattering properties depending on a K⁰ parameter and core radius is also investigated using two methods: Quantum Defect Theory and analytical formula of Symskoviski.     

Parallel-machine rescheduling with machine disruptions

In this study we consider a rescheduling problem on identical parallel machines. The rescheduling is undertaken because of a period of unavailability on one of the machines. We consider the total flow time as an efficiency measure and stability is gauged in terms of the number of jobs processed on different machines in the original and new schedules. We show that all efficient schedules with respect to efficiency and stability measures can be generated in polynomial time.     

Microstereophotolithography using a liquid crystal display as dynamic mask-generator

 A new laser microstereophotolithography process has been developed in our laboratory to manufacture three-dimensional parts with a high accuracy. As usual in laser stereophotolithography or laser microstereophotolithography, the part is manufactured layer by layer by a light-induced space-resolved polymerization. Until now, in all the already existing microstereophotolithography devices a layer is manufactured vector by vector, by moving the part beneath the initiating light source which remains motionless. We developed a simpler and easier process, in which we can manufacture an entire layer by irradiating its whole surface only once: we used a liquid crystal display as a dynamic generator of masks. In the device we set up, we need only one mobile element, the z translator, all the others are fixed. We manufactured several different 3D microparts, in particular a piece of bevel microgearing with helicoidal cogs, the volume of which is less than half a cubic millimetre. Received: 14 December 1995 / Accepted: 16 September 1996     

Optimal joint positions and stiffness distribution for minimum mass frames with damping constraints

Optimal design of frames including cross-sectional dimensions (size parameters) and rigid joint positions between beams (configuration parameters) is treated in the paper. The optimal design corresponds to a minimal mass structure with constraints set on damping capacity of free vibration modes. The sensitivity analysis of distinct as well as multiple frequencies is performed analytically using a complex variable sensitivity method. The linking process of size and configuration variables is applied to generate different classes of optimal designs. The numerical algorithm is based on quadratic approximation of the objective function and linear approximation of active constraints. The examples are provided for 2, 12, and 124 beam frames.Presented at WCSMO-2, Zakopane, Poland, May 26–30, 1997     

Three‐dimensional finite element computations for frictional contact problems with non‐associated sliding rule

This paper presents an algorithm for solving anisotropic frictional contact problems where the sliding rule is non‐associated.The algorithm is based on a variational formulation of the complex interface model that combine the classical unilateral contact law and an anisotropic friction model with a non‐associated slip rule. Both the friction condition and the sliding potential are elliptical and have the same principal axes but with different semi‐axes ratio. The frictional contact law and its inverse are derived from a single non‐differentiable scalar‐valued function, called a bi‐potential. The convexity properties of the bi‐potential permit to associate stationary principles with initial/boundary value problems. With the present formulation, the time‐integration of the frictional contact law takes the form of a projection onto a convex set and only one predictor–corrector step addresses all cases (sticking, sliding, no‐contact). A solution algorithm is presented and tested on a simple example that shows the strong influence of the slip rule on the frictional behaviour. Copyright 2004 John Wiley & Sons, Ltd.