Abstract: This paper aims to provide an interpretation of artificial neural networks (ANNs) and explore some of its implications. The interpretation views ANNs as a memory which encodes instances of experience. An experiment explores the behavior of encoding and retrieval of instances from memory. A localised representation ANN is created that allows control over encoding and retrieved memory sample size and is experimented with using the MNIST digits dataset. The relationship between input familiarity, conflict within retrieved samples, and error rates is described and demonstrated to be an effective driver for memory encoding. Results indicate that selective encoding and retrieval samples that allow detection of memory conflicts produce optimal performance, and that error rates are normally distributed with input familiarity and conflict. By using input familiarity and sample consistency to guide memory encoding, the number of encoding trials on the dataset were reduced to 18.33% of the training data while maintaining good recognition performance on the test data.
Abstract: The aim of this paper is to compare and discuss better classifier algorithm options for credit risk assessment by applying different Machine Learning techniques. Using records from a Brazilian financial institution, this study uses a database of 5,432 companies that are clients of the bank, where 2,600 clients are classified as non-defaulters, 1,551 are classified as defaulters and 1,281 are temporarily defaulters, meaning that the clients are overdue on their payments for up 180 days. For each case, a total of 15 attributes was considered for a one-against-all assessment using four different techniques: Artificial Neural Networks Multilayer Perceptron (ANN-MLP), Artificial Neural Networks Radial Basis Functions (ANN-RBF), Logistic Regression (LR) and finally Support Vector Machines (SVM). For each method, different parameters were analyzed in order to obtain different results when the best of each technique was compared. Initially the data were coded in thermometer code (numerical attributes) or dummy coding (for nominal attributes). The methods were then evaluated for each parameter and the best result of each technique was compared in terms of accuracy, false positives, false negatives, true positives and true negatives. This comparison showed that the best method, in terms of accuracy, was ANN-RBF (79.20% for non-defaulter classification, 97.74% for defaulters and 75.37% for the temporarily defaulter classification). However, the best accuracy does not always represent the best technique. For instance, on the classification of temporarily defaulters, this technique, in terms of false positives, was surpassed by SVM, which had the lowest rate (0.07%) of false positive classifications. All these intrinsic details are discussed considering the results found, and an overview of what was presented is shown in the conclusion of this study.
Abstract: The primary approach for estimating bridge deterioration uses Markov-chain models and regression analysis. Traditional Markov models have problems in estimating the required transition probabilities when a small sample size is used. Often, reliable bridge data have not been taken over large periods, thus large data sets may not be available. This study presents an important change to the traditional approach by using the Small Data Method to estimate transition probabilities. The results illustrate that the Small Data Method and traditional approach both provide similar estimates; however, the former method provides results that are more conservative. That is, Small Data Method provided slightly lower than expected bridge condition ratings compared with the traditional approach. Considering that bridges are critical infrastructures, the Small Data Method, which uses more information and provides more conservative estimates, may be more appropriate when the available sample size is small. In addition, regression analysis was used to calculate bridge deterioration. Condition ratings were determined for bridge groups, and the best regression model was selected for each group. The results obtained were very similar to those obtained when using Markov chains; however, it is desirable to use more data for better results.
Abstract: With the increasing use and application of Wireless Sensor Networks (WSN), need has arisen to explore them in more effective and efficient manner. An important area which can bring efficiency to WSNs is the localization process, which refers to the estimation of the position of wireless sensor nodes in an ad hoc network setting, in reference to a coordinate system that may be internal or external to the network. In this paper, we have done comparison and analysed Sigmoidal Feedforward Artificial Neural Networks (SFFANNs) and Radial Basis Function (RBF) networks for developing localization framework in WSNs. The presented work utilizes the Received Signal Strength Indicator (RSSI), measured by static node on 100 x 100 m2 grid from three anchor nodes. The comprehensive evaluation of these approaches is done using MATLAB software. The simulation results effectively demonstrate that FFANNs based sensor motes will show better localization accuracy as compared to RBF.
Abstract: There are about 1% of the world population suffering
from the hidden disability known as epilepsy and major developing
countries are not fully equipped to counter this problem. In order to
reduce the inconvenience and danger of epilepsy, different methods
have been researched by using a artificial neural network (ANN)
classification to distinguish epileptic waveforms from normal brain
waveforms. This paper outlines the aim of achieving massive
ANN parallelization through a dedicated hardware using bit-serial
processing. The design of this bit-serial Neural Processing Element
(NPE) is presented which implements the functionality of a complete
neuron using variable accuracy. The proposed design has been tested
taking into consideration non-idealities of a hardware ANN. The NPE
consists of a bit-serial multiplier which uses only 16 logic elements
on an Altera Cyclone IV FPGA and a bit-serial ALU as well as a
look-up table. Arrays of NPEs can be driven by a single controller
which executes the neural processing algorithm. In conclusion, the
proposed compact NPE design allows the construction of complex
hardware ANNs that can be implemented in a portable equipment
that suits the needs of a single epileptic patient in his or her daily
activities to predict the occurrences of impending tonic conic seizures.
Abstract: The Cone Penetration Test (CPT) is a common in-situ
test which generally investigates a much greater volume of soil more
quickly than possible from sampling and laboratory tests. Therefore,
it has the potential to realize both cost savings and assessment of soil
properties rapidly and continuously. The principle objective of this
paper is to demonstrate the feasibility and efficiency of using
artificial neural networks (ANNs) to predict the soil angle of internal
friction (Φ) and the soil modulus of elasticity (E) from CPT results
considering the uncertainties and non-linearities of the soil. In
addition, ANNs are used to study the influence of different
parameters and recommend which parameters should be included as
input parameters to improve the prediction. Neural networks discover
relationships in the input data sets through the iterative presentation
of the data and intrinsic mapping characteristics of neural topologies.
General Regression Neural Network (GRNN) is one of the powerful
neural network architectures which is utilized in this study. A large
amount of field and experimental data including CPT results, plate
load tests, direct shear box, grain size distribution and calculated data
of overburden pressure was obtained from a large project in the
United Arab Emirates. This data was used for the training and the
validation of the neural network. A comparison was made between
the obtained results from the ANN's approach, and some common
traditional correlations that predict Φ and E from CPT results with
respect to the actual results of the collected data. The results show
that the ANN is a very powerful tool. Very good agreement was
obtained between estimated results from ANN and actual measured
results with comparison to other correlations available in the
literature. The study recommends some easily available parameters
that should be included in the estimation of the soil properties to
improve the prediction models. It is shown that the use of friction
ration in the estimation of Φ and the use of fines content in the
estimation of E considerable improve the prediction models.
Abstract: Artificial neural networks (ANN) have the ability to model input-output relationships from processing raw data. This characteristic makes them invaluable in industry domains where such knowledge is scarce at best. In the recent decades, in order to overcome the black-box characteristic of ANNs, researchers have attempted to extract the knowledge embedded within ANNs in the form of rules that can be used in inference systems. This paper presents a new technique that is able to extract a small set of rules from a two-layer ANN. The extracted rules yield high classification accuracy when implemented within a fuzzy inference system. The technique targets industry domains that possess less complex problems for which no expert knowledge exists and for which a simpler solution is preferred to a complex one. The proposed technique is more efficient, simple, and applicable than most of the previously proposed techniques.
Abstract: HIV-1 genome is highly heterogeneous. Due to this
variation, features of HIV-I genome is in a wide range. For this
reason, the ability to infection of the virus changes depending on
different chemokine receptors. From this point of view, R5 HIV
viruses use CCR5 coreceptor while X4 viruses use CXCR5 and
R5X4 viruses can utilize both coreceptors. Recently, in
Bioinformatics, R5X4 viruses have been studied to classify by using
the experiments on HIV-1 genome.
In this study, R5X4 type of HIV viruses were classified using
Auto Regressive (AR) model through Artificial Neural Networks
(ANNs). The statistical data of R5X4, R5 and X4 viruses was
analyzed by using signal processing methods and ANNs. Accessible
residues of these virus sequences were obtained and modeled by AR
model since the dimension of residues is large and different from
each other. Finally the pre-processed data was used to evolve various
ANN structures for determining R5X4 viruses. Furthermore ROC
analysis was applied to ANNs to show their real performances. The
results indicate that R5X4 viruses successfully classified with high
sensitivity and specificity values training and testing ROC analysis
for RBF, which gives the best performance among ANN structures.
Abstract: .Hardware realization of a Neural Network (NN), to a large extent depends on the efficient implementation of a single neuron. FPGA-based reconfigurable computing architectures are suitable for hardware implementation of neural networks. FPGA realization of ANNs with a large number of neurons is still a challenging task. This paper discusses the issues involved in implementation of a multi-input neuron with linear/nonlinear excitation functions using FPGA. Implementation method with resource/speed tradeoff is proposed to handle signed decimal numbers. The VHDL coding developed is tested using Xilinx XC V50hq240 Chip. To improve the speed of operation a lookup table method is used. The problems involved in using a lookup table (LUT) for a nonlinear function is discussed. The percentage saving in resource and the improvement in speed with an LUT for a neuron is reported. An attempt is also made to derive a generalized formula for a multi-input neuron that facilitates to estimate approximately the total resource requirement and speed achievable for a given multilayer neural network. This facilitates the designer to choose the FPGA capacity for a given application. Using the proposed method of implementation a neural network based application, namely, a Space vector modulator for a vector-controlled drive is presented
Abstract: The quality of short term load forecasting can improve the efficiency of planning and operation of electric utilities. Artificial Neural Networks (ANNs) are employed for nonlinear short term load forecasting owing to their powerful nonlinear mapping capabilities. At present, there is no systematic methodology for optimal design and training of an artificial neural network. One has often to resort to the trial and error approach. This paper describes the process of developing three layer feed-forward large neural networks for short-term load forecasting and then presents a heuristic search algorithm for performing an important task of this process, i.e. optimal networks structure design. Particle Swarm Optimization (PSO) is used to develop the optimum large neural network structure and connecting weights for one-day ahead electric load forecasting problem. PSO is a novel random optimization method based on swarm intelligence, which has more powerful ability of global optimization. Employing PSO algorithms on the design and training of ANNs allows the ANN architecture and parameters to be easily optimized. The proposed method is applied to STLF of the local utility. Data are clustered due to the differences in their characteristics. Special days are extracted from the normal training sets and handled separately. In this way, a solution is provided for all load types, including working days and weekends and special days. The experimental results show that the proposed method optimized by PSO can quicken the learning speed of the network and improve the forecasting precision compared with the conventional Back Propagation (BP) method. Moreover, it is not only simple to calculate, but also practical and effective. Also, it provides a greater degree of accuracy in many cases and gives lower percent errors all the time for STLF problem compared to BP method. Thus, it can be applied to automatically design an optimal load forecaster based on historical data.
Abstract: In this paper, an Arabic letter recognition system based on Artificial Neural Networks (ANNs) and statistical analysis for feature extraction is presented. The ANN is trained using the Least Mean Squares (LMS) algorithm. In the proposed system, each typed Arabic letter is represented by a matrix of binary numbers that are used as input to a simple feature extraction system whose output, in addition to the input matrix, are fed to an ANN. Simulation results are provided and show that the proposed system always produces a lower Mean Squared Error (MSE) and higher success rates than the current ANN solutions.
Abstract: Designing and implementing intelligent systems has become a crucial factor for the innovation and development of better products of space technologies. A neural network is a parallel system, capable of resolving paradigms that linear computing cannot. Field programmable gate array (FPGA) is a digital device that owns reprogrammable properties and robust flexibility. For the neural network based instrument prototype in real time application, conventional specific VLSI neural chip design suffers the limitation in time and cost. With low precision artificial neural network design, FPGAs have higher speed and smaller size for real time application than the VLSI and DSP chips. So, many researchers have made great efforts on the realization of neural network (NN) using FPGA technique. In this paper, an introduction of ANN and FPGA technique are briefly shown. Also, Hardware Description Language (VHDL) code has been proposed to implement ANNs as well as to present simulation results with floating point arithmetic. Synthesis results for ANN controller are developed using Precision RTL. Proposed VHDL implementation creates a flexible, fast method and high degree of parallelism for implementing ANN. The implementation of multi-layer NN using lookup table LUT reduces the resource utilization for implementation and time for execution.
Abstract: Prediction of highly non linear behavior of suspended
sediment flow in rivers has prime importance in the field of water
resources engineering. In this study the predictive performance of
two Artificial Neural Networks (ANNs) namely, the Radial Basis
Function (RBF) Network and the Multi Layer Feed Forward (MLFF)
Network have been compared. Time series data of daily suspended
sediment discharge and water discharge at Pari River was used for
training and testing the networks. A number of statistical parameters
i.e. root mean square error (RMSE), mean absolute error (MAE),
coefficient of efficiency (CE) and coefficient of determination (R2)
were used for performance evaluation of the models. Both the models
produced satisfactory results and showed a good agreement between
the predicted and observed data. The RBF network model provided
slightly better results than the MLFF network model in predicting
suspended sediment discharge.
Abstract: We study the dynamic response of a wind turbine
structure subjected to theoretical seismic motions, taking into account
the rotational component of ground shaking. Models are generated
for a shallow moderate crustal earthquake in the Madrid Region
(Spain). Synthetic translational and rotational time histories are
computed using the Discrete Wavenumber Method, assuming a point
source and a horizontal layered earth structure. These are used to
analyze the dynamic response of a wind turbine, represented by a
simple finite element model. Von Mises stress values at different
heights of the tower are used to study the dynamical structural
response to a set of synthetic ground motion time histories
Abstract: Artificial Neural Networks (ANNs) have been used successfully in many scientific, industrial and business domains as a method for extracting knowledge from vast amounts of data. However the use of ANN techniques in the sporting domain has been limited. In professional sport, data is stored on many aspects of teams, games, training and players. Sporting organisations have begun to realise that there is a wealth of untapped knowledge contained in the data and there is great interest in techniques to utilise this data. This study will use player data from the elite Australian Football League (AFL) competition to train and test ANNs with the aim to predict the onset of injuries. The results demonstrate that an accuracy of 82.9% was achieved by the ANNs’ predictions across all examples with 94.5% of all injuries correctly predicted. These initial findings suggest that ANNs may have the potential to assist sporting clubs in the prediction of injuries.
Abstract: This paper is focused on issues of process modeling
and two model based control strategies of a fed-batch sugar
crystallization process applying the concept of artificial neural
networks (ANNs). The control objective is to force the operation into
following optimal supersaturation trajectory. It is achieved by
manipulating the feed flow rate of sugar liquor/syrup, considered as
the control input. The control task is rather challenging due to the
strong nonlinearity of the process dynamics and variations in the
crystallization kinetics. Two control alternatives are considered –
model predictive control (MPC) and feedback linearizing control
(FLC). Adequate ANN process models are first built as part of the
controller structures. MPC algorithm outperforms the FLC approach
with respect to satisfactory reference tracking and smooth control
action. However, the MPC is computationally much more involved
since it requires an online numerical optimization, while for the FLC
an analytical control solution was determined.
Abstract: Many artificial intelligence (AI) techniques are inspired
by problem-solving strategies found in nature. Robustness is a key
feature in many natural systems. This paper studies robustness in
artificial neural networks (ANNs) and proposes several novel, nature
inspired ANN architectures. The paper includes encouraging results
from experimental studies on these networks showing increased
robustness.
Abstract: The development of Artificial Neural Networks
(ANNs) is usually a slow process in which the human expert has to
test several architectures until he finds the one that achieves best
results to solve a certain problem. This work presents a new
technique that uses Genetic Programming (GP) for automatically
generating ANNs. To do this, the GP algorithm had to be changed in
order to work with graph structures, so ANNs can be developed. This
technique also allows the obtaining of simplified networks that solve
the problem with a small group of neurons. In order to measure the
performance of the system and to compare the results with other
ANN development methods by means of Evolutionary Computation
(EC) techniques, several tests were performed with problems based
on some of the most used test databases. The results of those
comparisons show that the system achieves good results comparable
with the already existing techniques and, in most of the cases, they
worked better than those techniques.
Abstract: Signature represents an individual characteristic of a
person which can be used for his / her validation. For such application
proper modeling is essential. Here we propose an offline signature
recognition and verification scheme which is based on extraction of
several features including one hybrid set from the input signature
and compare them with the already trained forms. Feature points
are classified using statistical parameters like mean and variance.
The scanned signature is normalized in slant using a very simple
algorithm with an intention to make the system robust which is
found to be very helpful. The slant correction is further aided by the
use of an Artificial Neural Network (ANN). The suggested scheme
discriminates between originals and forged signatures from simple
and random forgeries. The primary objective is to reduce the two
crucial parameters-False Acceptance Rate (FAR) and False Rejection
Rate (FRR) with lesser training time with an intension to make the
system dynamic using a cluster of ANNs forming a multiple classifier
system.
Abstract: this paper aims to provide an approach to predict the
performance of the product produced after multi-stages of
manufacturing processes, as well as the assembly. Such approach
aims to control and subsequently identify the relationship between
the process inputs and outputs so that a process engineer can more
accurately predict how the process output shall perform based on the
system inputs. The approach is guided by a six-sigma methodology to
obtain improved performance.
In this paper a case study of the manufacture of a hermetic
reciprocating compressor is presented. The application of artificial
neural networks (ANNs) technique is introduced to improve
performance prediction within this manufacturing environment. The
results demonstrate that the approach predicts accurately and
effectively.