Multi-Objective Feature Subset Selection using Non-dominated Sorting Genetic Algorithm

Journal of Applied Research and Technology, Journal Year: 2015, Volume and Issue: 13(1), P. 145 - 159

Published: Feb. 1, 2015

This paper presents an evolutionary algorithm based technique to solve multi-objective feature subset selection problem. The data used for classification contains large number of features called attributes. Some these attributes are not relevant and needs be eliminated. In procedure, each has effect on the accuracy, cost learning time classifier. So, there is a strong requirement select before building proposed treats as optimization research uses one latest genetic algorithms (NSGA - II). fitness value particular measured by using ID3. testing accuracy acquired then assigned value. tested several datasets taken from UCI machine repository. experiments demonstrate feasibility NSGA-II selection.

Language: Английский

---

FORECASTING GOLD PRICE CHANGES BY USING ADAPTIVE NETWORK FUZZY INFERENCE SYSTEM

Journal of Business Economics and Management, Journal Year: 2012, Volume and Issue: 13(5), P. 994 - 1010

Published: Oct. 4, 2012

Developing a precise and accurate model of gold price is critical to assets management because its unique features. In this paper, adaptive neuro-fuzzy inference system (ANFIS) artificial neural network (ANN) have been used for modeling the price, compared with traditional statistical ARIMA (autoregressive integrated moving average). The three performance measures, coefficient determination (R 2), root mean squared error (RMSE), absolute (MAE), are utilized evaluate performances different models developed. results show that ANFIS outperforms other (i.e. ANN model), in terms criteria during training validation phases. Sensitivity analysis showed changes highly dependent upon values silver oil price.

Language: Английский

---

Evolutionary fuzzy modeling

IEEE Transactions on Fuzzy Systems, Journal Year: 2003, Volume and Issue: 11(5), P. 652 - 665

Published: Oct. 1, 2003

This study is concerned with a general methodology of identification fuzzy models. Unlike numeric models, models operate at level information granules - sets and this aspect brings up an important design requirement transparency the model. We propose three-phase development framework by distinguishing between structural parametric optimization processes. The underlying topology model dwells on neural networks architectures governed logic equipped flexibility. Two mechanisms are explored: realized via genetic programming whereas for ensuing detailed we proceed gradient-based learning. main advantages approach discussed in detail. illustrated aid example that provides insight into performance quantifies crucial issues.

Language: Английский

---

Visualizing the computational intelligence field [Application Notes]

IEEE Computational Intelligence Magazine, Journal Year: 2006, Volume and Issue: 1(4), P. 6 - 10

Published: Nov. 1, 2006

In this paper, we visualize the structure and evolution of computational intelligence (CI) field. Based on our visualizations, analyze way in which CI field is divided into several subfields. The visualizations provide insight characteristics each subfield relations between By comparing two one based data from 2002 2006, examine how has evolved over last years. A quantitative analysis further identifies a number emerging areas within that use consist abstracts papers presented at IEEE World Congress Computational Intelligence (WCCI) 2006. Using fully automatic procedure, so-called concept maps are constructed data. These associations main concepts Our largely

Language: Английский

---

Have artificial neural networks met expectations in drug discovery as implemented in QSAR framework?

Expert Opinion on Drug Discovery, Journal Year: 2016, Volume and Issue: 11(7), P. 627 - 639

Published: May 5, 2016

Artificial neural networks (ANNs) are highly adaptive nonlinear optimization algorithms that have been applied in many diverse scientific endeavors, ranging from economics, engineering, physics, and chemistry to medical science. Notably, the past two decades, ANNs used widely process of drug discovery.In this review, authors discuss advantages disadvantages discovery as incorporated into quantitative structure-activity relationships (QSAR) framework. Furthermore, examine recent studies, which span over a broad area with various diseases discovery. In addition, attempt answer question about expectations trends field.The old pitfalls overtraining interpretability still present ANNs. However, despite these pitfalls, believe likely met researchers considered excellent tools for data modeling QSAR. It is will continue be development future.

Language: Английский

---

Comparing PSO structures to learn the game of checkers from zero knowledge

Published: July 9, 2004

This paper investigates the effectiveness of various particle swarm optimiser structures to learn how play game checkers. Co-evolutionary techniques are used train playing agents. Performance is compared against a player making moves at random. Initial experimental results indicate definite advantages in using certain information sharing and size configurations successfully

Language: Английский

---

Cosmological parameter estimation using particle swarm optimization

Physical review. D. Particles, fields, gravitation, and cosmology/Physical review. D, Particles, fields, gravitation, and cosmology, Journal Year: 2012, Volume and Issue: 85(12)

Published: June 19, 2012

Constraining theoretical models, which are represented by a set of parameters, using observational data is an important exercise in cosmology. In Bayesian framework this done finding the probability distribution parameters best fits to sampling based methods like Markov chain Monte Carlo (MCMC). It has been argued that MCMC may not be option certain problems target function (likelihood) poses local maxima or have very high dimensionality. Apart from this, there examples we mainly interested find point parameter space at largest value. situation problem estimation becomes optimization problem. present work show particle swarm (PSO), artificial intelligence inspired population search procedure, can also used for cosmological estimation. Using PSO were able recover best-fit $\ensuremath{\Lambda}$ cold dark matter (LCDM) model WMAP seven year without any prior guess value other property standard deviation, as common MCMC. We report results consider binned primordial power spectrum (to increase dimensionality problem) and with features gives lower chi square than law. Since does sample likelihood surface fair way, follow fitting procedure spread around point.

Language: Английский

---

Flocking based approach for data clustering

Natural Computing, Journal Year: 2009, Volume and Issue: 9(3), P. 767 - 791

Published: Dec. 23, 2009

Language: Английский

---

Design of very thin wide band absorbers using modified local best particle swarm optimization

AEU - International Journal of Electronics and Communications, Journal Year: 2007, Volume and Issue: 62(7), P. 549 - 556

Published: July 25, 2007

Language: Английский

---

Measuring flow as concept for detecting game fun in the Pac-Man game

Published: June 1, 2008

Popular games often have a high-quality graphic design but quite simple-minded non player characters (NPC). Recently, Computational Intelligence (CI) methods been discovered as suitable to revive NPC, making more interesting, challenging, and funny. We present fairly large study of human players on the simple arcade game Pac-Man, controlling ghosts behaviors by strategies, neural networks or evolutionary algorithms. The playerpsilas fun is course subjective experience, we presume that it related psychological flow concept. deal with question whether reliable measure than asking directly for experienced during game. In order detect flow, introduce based interaction time fraction between human-controlled Pac-Man ghosts, compare outcome results suggested Yannakakis Hallam [1].

Language: Английский

---