A novel approach to multiparametric quadratic programming

Multiparametric (mp) programming pre-computes optimal solutions offline which are functions of parameters whose values become apparent online. This makes it particularly well suited for applications that need a rapid solution of online optimization problems. In this work, we propose a novel approach to multiparametric programming problems based on an enumeration of active sets and use it to obtain a parametric solution for a convex quadratic program (QP). To avoid the combinatorial explosion of the enumeration procedure, an active set pruning criterion is presented that makes the enumeration implicit. The method guarantees that all regions of the partition are critical regions without any artificial cuts, and further that no region of the parameter space is left unexplored.     

Regularization approach to inductive genetic programming

This paper presents an approach to regularization of inductive genetic programming tuned for learning polynomials. The objective is to achieve optimal evolutionary performance when searching high-order multivariate polynomials represented as tree structures. We show how to improve the genetic programming of polynomials by balancing its statistical bias with its variance. Bias reduction is achieved by employing a set of basis polynomials in the tree nodes for better agreement with the examples. Since this often leads to over-fitting, such tendencies are counteracted by decreasing the variance through regularization of the fitness function. We demonstrate that this balance facilitates the search as well as enables discovery of parsimonious, accurate, and predictive polynomials. The experimental results given show that this regularization approach outperforms traditional genetic programming on benchmark data mining and practical time-series prediction tasks     

A genetic encoding approach for learning methods for combining classifiers

Several studies have reported that the ensemble of classifiers can improve the performance of a stand-alone classifier. In this paper, we propose a learning method for combining the predictions of a set of classifiers.The method described in this paper uses a genetic-based version of the correspondence analysis for combining classifiers. The correspondence analysis is based on the orthonormal representation of the labels assigned to the patterns by a pool of classifiers. In this paper instead of the orthonormal representation we use a pool of representations obtained by a genetic algorithm. Each single representation is used to train a different classifiers, these classifiers are combined by vote rule.The performance improvement with respect to other learning-based fusion methods is validated through experiments with several benchmark datasets.     

Morphological algorithm design for binary images using genetic programming

This paper presents a Genetic Programming (GP) approach to the design of Mathematical Morphology (MM) algorithms for binary images. The algorithms are constructed using logic operators and the basic MM operators, i.e. erosion and dilation, with a variety of structuring elements. GP is used to evolve MM algorithms that convert a binary image into another containing just a particular feature of interest. In the study we have tested three fitness functions, training sets with different numbers of elements, training images of different sizes, and 7 different features in two different kinds of applications. The results obtained show that it is possible to evolve good MM algorithms using GP.     

A novel approach for automated land partitioning using genetic algorithm

Land consolidation is an important tool to prevent land fragmentation and enhance agricultural productivity. Land partitioning is one of the most significant problems within the land consolidation process. This process is related to the subdivision of a block having non-uniform geometric shapes. Land partitioning determines the location of new land parcels and is a complex problem containing many conflicting demands, so conventional programming techniques are not sufficient for this NP optimization problem. Therefore, it is necessary to have an intelligent system with a standard decision-making mechanism capable of processing many criteria simultaneously and evaluating a number of different solutions in a short time. To overcome this problem and accelerate the land partitioning process, we proposed automated land partitioning using a genetic algorithm (ALP-GA). Besides the parcel's size, shape and land value, the proposed method evaluates fixed facilities, and the degree and location of cadastral parcels to generate a land partitioning plan. The proposed method automated the land partitioning process using an intelligent system and was implemented over a real project area. Experimental study shows that the proposed method is more successful and efficient than the designer with respect to the results meeting the objective function. In addition, the land partition process is greatly simplified by the proposed method.     

A functional approach to generic programming using adaptive traversals

Writing functions over complex user-defined datatypes can be tedious and error prone. Generic (or polytypic) programming and higher order functions like foldr have resolved some of these issues, but can be too general to be practically useful for larger collections of data types. In this paper we present a traversal-based approach to generic programming using function sets. Our traversal is an adaptive, higher-order function that employs an asymmetric type-based multiple dispatch to fold over arbitrarily complex structures. We introduce our approach in the context of our Scheme library implementation, present a typed model of our system, and provide a proof of type soundness, showing that our flexible, adaptive approach is both useful and safe.     

A hierarchical approach to multi-class fuzzy classifiers

In this paper we present a hierarchical approach for generating fuzzy rules directly from data in a simple and effective way. The fuzzy classifier results from the union of fuzzy systems, employing the Wang and Mendel algorithm, built on input regions increasingly smaller, according to a multi-level grid-like partition. Key parameters of the proposed method are optimized by means of a genetic algorithm. Only the necessary partitions are built, in order to guarantee high interpretability and to avoid the explosion of the number of rules as the hierarchical level increases. We apply our method to real-world data collected from a photovoltaic (PV) installation so as to linguistically describe how the temperature of the PV panel and the irradiation relate to the class (low, medium, high) of the energy produced by the panel. The obtained mean and maximum classification percentages on 30 repetitions of the experiment are 97.38% and 97.91%, respectively. We also apply our method to the classification of some well-known benchmark datasets and show how the achieved results compare favourably with those obtained by other authors using different techniques.     

A novel approach to discrete truss design problems using mixed integer neighborhood search

A new approach for the multi-objective optimization of composite structures under the effects of uncertainty in mechanical properties, structural parameters and external loads is proposed, to guarantee higher levels of accuracy exclusively with Evolutionary Algorithms (EA). The concepts of Reliability-Based Robust Design Optimization (RBRDO) are applied. Optimality is defined as the minimization of the structural weight and robustness as the minimization of the determinant of the variance-covariance matrix of the structural responses. Reliability assessment is performed through a mathematical reformulation of the Performance Measure Approach, suitable for EA, where the standard normal uncertainty space was defined in directional coordinates and reduced to the surface of the hypersphere of radius β^^a. A binary reliability constraint, that allowed avoiding unnecessary runs of the reliability inner-cycle is defined. The Robust Design Optimization cycle is solved by a multi-objective EA, based on constrained-dominance. Sensitivities of the structural responses, necessary for uncertainty analysis only, are calculated analytically by the Adjoint Variable Method. A numerical example considering a balanced angle-ply laminate shell is presented. Results show an effective convergence of the Pareto-optimal Front (POF). Uncertainty analysis shows that the variability of the critical displacements increases along the POF. For the stresses, variability is stable but of higher values. The incorporation of the reliability constraint prevents the natural decrease of the reliability index, along the POF, to reach levels too close, or inside, of the failure domain. The distribution of the reliability measures along the POF is similar and demonstrates the effects of reliability in the RBRDO procedure.     

A semi-dependent decomposition approach to learn hierarchical classifiers

In hierarchical classification, classes are arranged in a hierarchy represented by a tree or a forest, and each example is labeled with a set of classes located on paths from roots to leaves or internal nodes. In other words, both multiple and partial paths are allowed. A straightforward approach to learn a hierarchical classifier, usually used as a baseline method, consists in learning one binary classifier for each node of the hierarchy; the hierarchical classifier is then obtained using a top-down evaluation procedure. The main drawback of this naive approach is that these binary classifiers are constructed independently, when it is clear that there are dependencies between them that are motivated by the hierarchy and the evaluation procedure employed. In this paper, we present a new decomposition method in which each node classifier is built taking into account other classifiers, its descendants, and the loss function used to measure the goodness of hierarchical classifiers. Following a bottom-up learning strategy, the idea is to optimize the loss function at every subtree assuming that all classifiers are known except the one at the root. Experimental results show that the proposed approach has accuracies comparable to state-of-the-art hierarchical algorithms and is better than the naive baseline method described above. Moreover, the benefits of our proposal include the possibility of parallel implementations, as well as the use of all available well-known techniques to tune binary classification SVMs.     

A modular approach to real-time programming using actors and Java

This paper describes an actor-based approach to real-time programming, which focuses on the separation of functional from timing behaviour. The approach favours modularity and time predictability. Clusters of actors, allocated on distinct processors, are orchestrated by a control machine which provides an event-driven and time-driven customisable scheduling framework. The approach can be hosted by Java, which fosters a clean and type-safe programming style. Temporal analysis can be formally assisted by Coloured Petri Nets.     

A controlled genetic programming approach for the deceptive domain

Traditional genetic programming (GP) randomly combines subtrees by applying crossover. There is a growing interest in methods that can control such recombination operations in order to achieve faster convergence. In this paper, a new approach is presented for guiding the recombination process for genetic programming. The method is based on extracting the global information of the promising solutions that appear during the genetic search. The aim is to use this information to control the crossover operation afterwards. A separate control module is used to process the collected information. This module guides the search process by sending feedback to the genetic engine about the consequences of possible recombination alternatives.     

A regression tree approach using mathematical programming

Regression analysis is a machine learning approach that aims to accurately predict the value of continuous output variables from certain independent input variables, via automatic estimation of their latent relationship from data. Tree-based regression models are popular in literature due to their flexibility to model higher order non-linearity and great interpretability. Conventionally, regression tree models are trained in a two-stage procedure, i.e. recursive binary partitioning is employed to produce a tree structure, followed by a pruning process of removing insignificant leaves, with the possibility of assigning multivariate functions to terminal leaves to improve generalisation. This work introduces a novel methodology of node partitioning which, in a single optimisation model, simultaneously performs the two tasks of identifying the break-point of a binary split and assignment of multivariate functions to either leaf, thus leading to an efficient regression tree model. Using six real world benchmark problems, we demonstrate that the proposed method consistently outperforms a number of state-of-the-art regression tree models and methods based on other techniques, with an average improvement of 7–60% on the mean absolute errors (MAE) of the predictions.     

A genetic approach to automate preliminary design of gear drives

Determination of volume or weight of a gearbox is an important issue in preliminary design of power transmission applications. Trial and error procedure or some gear standards information sheets are commonly used in traditional design. The purpose of this paper is to automate preliminary design of gear drives by minimizing volume of gear trains. A stochastic approach Genetic Algorithm (GA) was applied to a parallel axis two stage helical gear trains problem. Static and dynamic penalty functions were introduced to the objective function for handling the design constraints. The results were compared with a deterministic design procedure developed. GA based approach produced quite well results promptly supplying preliminary design parameters of gear drives for different gear ratios to the designer.     

Evolutionary-computer-assisted design of image operators that detect interest points using genetic programming

This work describes a way of designing interest point detectors using an evolutionary-computer-assisted design approach. Nowadays, feature extraction is performed through the paradigm of interest point detection due to its simplicity and robustness for practical applications such as: image matching and view-based object recognition. Genetic programming is used as the core functionality of the proposed human-computer framework that significantly augments the scope of interest point design through a computer assisted learning process. Indeed, genetic programming has produced numerous interest point operators, many with unique or unorthodox designs. The analysis of those best detectors gives us an advantage to achieve a new level of creative design that improves the perspective for human-machine innovation. In particular, we present two novel interest point detectors produced through the analysis of multiple solutions that were obtained through single and multi-objective searches. Experimental results using a well-known testbed are provided to illustrate the performance of the operators and hence the effectiveness of the proposal.     

A novel approach to design optimal 2-D digital differentiator using vortex search optimization algorithm

Multimedia Tools and Applications - The designing of 2-D digital differentiator is multimodal and high dimensional problem which requires large number of differentiator coefficients to be...     

Automated passive filter synthesis using a novel tree representation and genetic programming

This paper proposes a novel tree representation which is suitable for the analysis of RLC (i.e., resistor, inductor, and capacitor) circuits. Genetic programming (GP) based on the tree representation is applied to passive filter synthesis problems. The GP is optimized and then incorporated into an algorithm which can automatically find parsimonious solutions without predetermining the number of the required circuit components. The experimental results show the proposed method is efficient in three aspects. First, the GP-evolved circuits are more parsimonious than those resulting from traditional design methods in many cases. Second, the proposed method is faster than previous work and can effectively generate parsimonious filters of very high order where conventional methods fail. Third, when the component values are restricted to a set of preferred values, the GP method can generate compliant solutions by means of novel circuit topology.     

A novel nonlinear programming approach for estimating CAPM beta of an asset using fuzzy regression

The evaluation of risky assets is one of the major research tasks in the finance theory. There are several Capital Asset Pricing Models (CAPM) in the literature; the most popular one of those is the Sharpe–Lintner–Black mean–variance CAPM. According to this model, the typical measure of systematic risk is the beta coefficient. The beta coefficient can be evaluated by means of least squares method (LSM), Robust Regression Techniques (RRT), or similar approaches. However, the statistical assumptions of LSM might be invalid in the existence of extreme observations in data set. In order to decrease influence on the beta coefficient of extreme observations, most analyst apply to RRT’s. However, either RRT’s remove the extreme observations from the data set, or decrease their influences on the beta coefficient. Whereas the omitted observations might be valuable for investors since they carry substantial information about the state of nature. In other words, there is a clash between statistical and financial theory. In this study, to overcome this incompatibility, and to take into account the extreme observations carried worthy information, a novel fuzzy regression approach is proposed. The proposed approach is based on both possibility concepts and central tendency in the estimation of beta coefficient. In application section of this paper, the beta coefficients of three assets traded in Istanbul Stock Exchange (ISE) are estimated by the proposed fuzzy approach and the traditional techniques, and then the results of analysis are compared, and discussed.     

A novel approach to the design of unknown input observers

A novel state estimator design scheme for linear dynamical systems driven by partially unknown inputs is presented. It is assumed that there is no information available about the unknown inputs, and thus no prior assumption is made about the nature of these inputs. A simple approach for designing a reduced-order unknown input observer (UIO) with pole-placement capability is proposed. By carefully examining the dynamic system involved and simple algebraic manipulations, it is possible to rewrite equations eliminating the unknown inputs from part of the system and to put them into a form where it could be partitioned into two interconnected subsystems, one of which is directly driven by known inputs only. This makes it possible to use a conventional Luenberger observer with a slight modification for the purpose of estimating the state of the system. As a result, it is also possible to state similar necessary and sufficient conditions to those of a conventional observer for the existence of a stable estimator and also arbitrary placement of the eigenvalues of the observer. The design and computational complexities involved in designing UIOs are greatly reduced in the proposed approach     

A novel approach to the explicit pole assignment self-tuningcontroller design

A recursive algorithm for the controller design phase of the adaptive pole assignment controller is presented. Compared to other pole-assignment adaptive controllers, the computation time to get the controller parameters is drastically reduced and the numerical stability of the controller parameters is increased. If persistent excitation is imposed, the system poles will be located at the desired position in the steady state and the global stability of the system can be easily obtained