-*- mode: org -*-
Supportive uses in general method classes for other areas

*  72(<-  3): Closed-Loop Restoration Approach to Blurry Images Based on Machine Learning and Feedback Optimization

Blind image deconvolution (BID) aims to remove or reduce the
degradations that have occurred during the acquisition or
processing. It is a challenging ill-posed problem due to a lack of
enough information in degraded image for unambiguous recovery of both
point spread function (PSF) and clear image. Although recently many
powerful algorithms appeared; however, it is still an active research
area due to the diversity of degraded images as well as
degradations. Closed-loop control systems are characterized with their
powerful ability to stabilize the behavior response and overcome
external disturbances by designing an effective feedback
optimization. In this paper, we employed feedback control to enhance
the stability of BID by driving the current estimation quality of PSF
to the desired level without manually selecting restoration parameters
and using an effective combination of machine learning with feedback
optimization. The foremost challenge when designing a feedback
structure is to construct or choose a suitable performance metric as a
controlled index and a feedback information. Our proposed quality
metric is based on the blur assessment of deconvolved patches to
identify the best PSF and computing its relative quality. The Kalman
filter-based extremum seeking approach is employed to find the optimum
value of controlled variable. To find better restoration parameters,
learning algorithms, such as multilayer perceptron and bagged decision
trees, are used to estimate the generic PSF support size instead of
trial and error methods. The problem is modeled as a combination of
pattern classification and regression using multiple training
features, including noise metrics, blur metrics, and low-level
statistics. Multi-objective genetic algorithm is used to find key
patches from multiple saliency maps which enhance performance and save
extra computation by avoiding ineffectual regions of the image. The
proposed scheme is shown to outperform corresponding open-loop
schemes, which often fails or needs many assumptions regarding images
and thus resulting in sub-optimal results.

2015



*  78(<-307): A note on the estimation of the Pareto efficient set for multiobjective matrix permutation problems

There are a number of important problems in quantitative psychology
that require the identification of a permutation of the n rows and
columns of an n x n proximity matrix. These problems encompass
applications such as unidimensional scaling, paired-comparison
ranking, and anti-Robinson forms. The importance of simultaneously
incorporating multiple objective criteria in matrix permutation
applications is well recognized in the literature; however, to date,
there has been a reliance on weighted-sum approaches that transform
the multiobjective problem into a single-objective optimization
problem. Although exact solutions to these single-objective problems
produce supported Pareto efficient solutions to the multiobjective
problem, many interesting unsupported Pareto efficient solutions may
be missed. We illustrate the limitation of the weighted-sum approach
with an example from the psychological literature and devise an
effective heuristic algorithm for estimating both the supported and
unsupported solutions of the Pareto efficient set.

2012



* 212(<-557): A neural stochastic multiscale optimization framework for sensor-based parameter estimation

This work presents a novel neural stochastic optimization framework
for reservoir parameter estimation that combines two independent
sources of spatial and temporal data: oil production data and dynamic
sensor data of flow pressures and concentrations. A parameter
estimation procedure is realized by minimizing a multi-objective
mismatch function between observed and predicted data. In order to be
able to efficiently perform large-scale parameter estimations, the
parameter space is decomposed in different resolution levels by means
of the singular value decomposition (SVD) and a wavelet upscaling
process. The estimation is carried out incrementally from low to
higher resolution levels by means of a neural stochastic multilevel
optimization approach. At a given resolution level, the parameter
space is globally explored and sampled by the simultaneous
perturbation stochastic approximation (SPSA) algorithm. The sampling
yielded by SPSA serves as training points for an artificial neural
network that allows for evaluating the sensitivity of different
multi-objective function components with respect to the model
parameters. The proposed approach may be suitable for different
engineering and scientific applications wherever the parameter space
results from discretizing a set of partial differential equations on a
given spatial domain.

2007


* 220(<-537): GIS and neural network method for potential road identification

Global positioning system (GPS)-based vehicle tracking systems were
used to track 20 vehicles involved in an 8-day field training exercise
at Yakima Training Center, Washington. A 3-layer feed-forward
artificial neural network (NN) with a backpropagation learning
algorithm was developed to identify potential roads. The NN was
trained using a subset of the GPS data that was supplemented with
field observations that documented newly formed road segments
resulting from concentrated vehicle traffic during the military
training exercise. The NN was subsequently applied to the full vehicle
movement data set to predict potential roads for the entire training
exercise. Model predictions were validated using additional
installation and site visit data. The first validation used the NN to
identify the existing road network as represented in the Yakima
Training Center GIS roads data layer Next, the NN was used to predict
emerging road networks that had not previously existed. The NN method
accurately classified approximately 94% of the training data, 85% of
the on-road movement data, and 78% of potential roads. The proposed NN
method more accurately classified potential roads than the previously
used multicriteria method, which was able to identify 10 out of 17
potential road segments across the entire training center.

2007


* 237(<-445): Photochemistry and chemometrics-An overview

Photochemistry has made significant contributions to our understanding
of many important natural processes as well as the scientific
discoveries of the man-made world. The measurements from such studies
are often complex and may require advanced data interpretation with
the use of multivariate or chemometrics methods. In general, such
methods have been applied successfully for data display,
classification, multivariate curve resolution and prediction in
analytical chemistry, environmental chemistry, engineering, medical
research and industry. However, in photochemistry, by comparison,
applications of such multivariate approaches were found to be less
frequent although a variety of methods have been used, especially with
spectroscopic photochemical applications. The methods include
Principal Component Analysis (PCA; data display), Partial Least
Squares (PLS; prediction), Artificial Neural Networks (ANN;
prediction) and several models for multivariate curve resolution
related to Parallel Factor Analysis (PARAFAC; decomposition of complex
responses). Applications of such methods are discussed in this
overview and typical examples include photodegradation of herbicides,
prediction of antibiotics in human fluids (fluorescence spectroscopy),
non-destructive in- and on-line monitoring (near infrared
spectroscopy) and fast-time resolution of spectroscopic signals from
photochemical reactions. It is also quite clear from the literature
that the scope of spectroscopic photochemistry was enhanced by the
application of chemometrics. To highlight and encourage further
applications of chemometrics in photochemistry, several additional
chemometrics approaches are discussed using data collected by the
authors. The use of a PCA biplot is illustrated with an analysis of a
matrix containing data on the performance of photocatalysts developed
for water splitting and hydrogen production. In addition, the
applications of the Multi-Criteria Decision Making (MCDM) ranking
methods and Fuzzy Clustering are demonstrated with an analysis of
water quality data matrix. Other examples of topics include the
application of simultaneous kinetic spectroscopic methods for
prediction of pesticides, and the use of response fingerprinting
approach for classification of medicinal preparations. In general, the
overview endeavours to emphasise the advantages of chemometrics'
interpretation of multivariate photochemical data, and an Appendix of
references and summaries of common and less usual chemometrics methods
noted in this work, is provided. Crown Copyright (C) 2010 Published by
Elsevier B.V. All rights reserved.

2009


* 240(<-146): Genetic Algorithm Based Multi-Objective Least Square Support Vector Machine for Simultaneous Determination of Multiple Components by Near Infrared Spectroscopy

The near infrared (NIR) spectrum contains a global signature of
composition, and enables to predict different properties of the
material. In the present paper, a genetic algorithm and an adaptive
modeling technique were applied to build a multi-objective least
square support vector machine (MLS-SVM), which was intended to
simultaneously determine the concentrations of multiple components by
NIR spectroscopy. Both the benchmark corn dataset and self-made
Forsythia suspense dataset were used to test the proposed
approach. Results show that a genetic algorithm combined with adaptive
modeling allows to efficiently search the LS-SVM hyperparameter
space. For the corn data, the performance of multi-objective LS-SVM
was significantly better than models built with PLS1 and PLS2
algorithms. As for the Forsythia suspense data, the performance of
multi-objective LS-SVM was equivalent to PLS1 and PLS2 models. In both
datasets, the over-fitting phenomena were observed on RBFNN
models. The single objective LS-SVM and MLS-SVM didn't show much
difference, but the one-time modeling convenience allows the potential
application of MLS-SVM to multicomponent NIR analysis.

2014


* 313(<-468): Signal denoising in engineering problems through the minimum gradient method

This paper applies the minimum gradient method (MGM) to denoise
signals in engineering problems. The MGM is a novel technique based on
the complexity control, which defines the learning as a bi-objective
problem in such a way to find the best trade-off between the empirical
risk and the machine complexity. A neural network trained with this
method can be used to pre-process data aiming at increasing the
signal-to-noise ratio (SNR). After training, the neural network
behaves as an adaptive filter which minimizes the cross-validation
error. By applying the general singular value decomposition (GSVD), we
show the relation between the proposed approach and the Wiener
filter. Some results are presented, including a toy example and two
complex engineering problems, which prove the effectiveness of the
proposed approach. (C) 2009 Elsevier B.V. All rights reserved.

2009



* 337(<-208): Evolutionary Surrogate Optimization of an Industrial Sintering Process

Despite showing immense potential as an optimization technique for
solving complex industrial problems, the use of evolutionary
algorithms, especially genetic algorithms (GAs), is restricted to
offline applications in many industrial cases due to their
computationally expensive nature. This problem becomes even more
severe when the underlying function as well as constraint evaluation
is computationally expensive. To reduce the overall application time
under this kind of scenario, a combined usage of the original
expensive model and a relatively less expensive surrogate model built
around the data provided by the original model in the course of
optimization has been proposed in this work. Use of surrogates
provides the quickness in the application, thereby saving the
execution time, and the use of original model allows the optimization
tool to be in the right path of the search process. Switching to the
surrogate model happens if predictability of the model is of
acceptable accuracy (to be decided by the decision maker), and thereby
the optimization time is saved without compromising the solution
quality. This concept of successive use of surrogate (artificial
neural network [ANN]) and original expensive model is applied on an
industrial two-layer sintering process where optimization decides the
individual thickness and coke content of each layer to maximize sinter
quality and minimize coke consumption simultaneously. The use of
surrogate could reduce the execution time by 60% and thereby improve
the decision support system utilization without compromising the
solution quality.

2013



* 338(<-344): Successive approximate model based multi-objective optimization for an industrial straight grate iron ore induration process using evolutionary algorithm

Multi-objective optimization of any complex industrial process using
first principle computationally expensive models often demands a
substantially higher computation time for evolutionary algorithms
making it less amenable for real time implementation. A combination of
the above-mentioned first principle model and approximate models based
on artificial neural network (ANN) successively learnt in due course
of optimization using the data obtained from first principle models
can be intelligently used for function evaluation and there by reduce
the aforementioned computational burden to a large extent. In this
work, a multi-objective optimization task (simultaneous maximization
of throughput and Tumble index) of an industrial iron ore induration
process has been studied to improve the operation of the process using
the above-mentioned metamodeling approach. Different pressure and
temperature values at different points of the furnace bed, grate speed
and bed height have been used as decision variables where as the
bounds on cold compression strength, abrasion index, maximum pellet
temperature and burn-through point temperature have been treated as
constraints. A popular evolutionary multi-objective algorithm, NSGA
II, amalgamated with the first principle model of the induration
process and its successively improving approximation model based on
ANN, has been adopted to carryout the task. The optimization results
show that as compared to the PO solutions obtained using only the
first principle model, (i) similar or better quality PO solutions can
be achieved by this metamodeling procedure with a close to 50% savings
in function evaluation and there by computation time and (ii) by
keeping the total number of function evaluations same, better quality
PO solutions can be obtained. (C) 2011 Elsevier Ltd. All rights
reserved.

2011


* 448(<- 96): High-level design space exploration of locally linear neuro-fuzzy models for embedded systems

Recently, artificial neural networks and neuro-fuzzy systems are being
introduced in embedded systems due to often-used solution for
classification and nonlinear system identification. In this paper, we
present a parametric neuro-fuzzy hardware and a framework for
exploring design space for an efficient hardware realization of
neuro-fuzzy models for embedded systems. The proposed hardware can be
used as a stand-alone core or be coupled with a central processing
unit for the purpose of online training. We also present a framework
to explore the design space for optimal design parameters (hardware
core parameters) so that an efficient neuro-fuzzy hardware in terms of
area, power consumption, and performance (delay) can be selected. It
examines whole design space to find Pareto efficient hardware without
increasing time-to-market and non-recurring engineering cost with the
aid of high-level design space exploration. Also, the performance of
the proposed hardware is compared against a soft core embedded
processor named NIOS II/s. The results obtained show that the selected
core is able to perform actions faster than NIOS II while it
dissipates less power. Moreover, the proposed framework is put into
action through three different scenarios to show off the capabilities
of framework for generating Pareto optimal points upon different
application demands. (C) 2013 Elsevier B.V. All rights reserved.

2014