Learning in Artificial Neural Networks: A Statistical Perspective

1989 ◽  
Vol 1 (4) ◽  
pp. 425-464 ◽  
Author(s):  
Halbert White
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Optimization Theory ◽  
Comprehensive Treatment ◽  
Training Methods ◽  
Network Learning ◽  
Advantages And Disadvantages ◽  
Systems Identification ◽  
Artificial Neural ◽  
Insight Into

The premise of this article is that learning procedures used to train artificial neural networks are inherently statistical techniques. It follows that statistical theory can provide considerable insight into the properties, advantages, and disadvantages of different network learning methods. We review concepts and analytical results from the literatures of mathematical statistics, econometrics, systems identification, and optimization theory relevant to the analysis of learning in artificial neural networks. Because of the considerable variety of available learning procedures and necessary limitations of space, we cannot provide a comprehensive treatment. Our focus is primarily on learning procedures for feedforward networks. However, many of the concepts and issues arising in this framework are also quite broadly relevant to other network learning paradigms. In addition to providing useful insights, the material reviewed here suggests some potentially useful new training methods for artificial neural networks.

A Survey on Various Applications of Artificial Neural Networks in Selected Fields of Healthcare

2011 ◽  
pp. 20-59
Author(s):  
Wolfgang I. Schollhorn ◽  
Jörg M. Jager
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Neuromuscular Control ◽  
Neural Nets ◽  
Biomedical Signals ◽  
Potential Benefits ◽  
Artificial Neural ◽  
Number Of Publications ◽  
Insight Into ◽  
Profound Insight

This chapter gives an overview of artificial neural networks as instruments for processing miscellaneous biomedical signals. A variety of applications are illustrated in several areas of healthcare. The structure of this chapter is rather oriented on medical fields like cardiology, gynecology, or neuromuscular control than on types of neural nets. Many examples demonstrate how neural nets can support the diagnosis and prediction of diseases. However, their content does not claim completeness due to the enormous amount and exponentially increasing number of publications in this field. Besides the potential benefits for healthcare, some remarks on underlying assumptions are also included as well as problems which may occur while applying artificial neural nets. It is hoped that this review gives profound insight into strengths as well as weaknesses of artificial neural networks as tools for processing biomedical signals.

MODULAR NEURAL NETWORKS: A SURVEY

1999 ◽  
Vol 09 (02) ◽  
pp. 129-151 ◽  
Author(s):  
GASSER AUDA ◽  
MOHAMED KAMEL
Keyword(s):  
Neural Networks ◽  
Decision Making ◽  
Artificial Neural Networks ◽  
Decomposition Techniques ◽  
Task Decomposition ◽  
Modular Neural Networks ◽  
Advantages And Disadvantages ◽  
Artificial Neural ◽  
Learning Schemes

Modular Neural Networks (MNNs) is a rapidly growing field in artificial Neural Networks (NNs) research. This paper surveys the different motivations for creating MNNs: biological, psychological, hardware, and computational. Then, the general stages of MNN design are outlined and surveyed as well, viz., task decomposition techniques, learning schemes and multi-module decision-making strategies. Advantages and disadvantages of the surveyed methods are pointed out, and an assessment with respect to practical potential is provided. Finally, some general recommendations for future designs are presented.

THE SYSTEM OF AUTOMATED DEVELOPMENT, LEARNING AND EXECUTION OF ARTIFICIAL NEURAL NETWORKS

2019 ◽  
pp. 100-107
Author(s):  
V.A. Sobolevsky
Keyword(s):  
Genetic Algorithm ◽  
Neural Networks ◽  
Artificial Neural Networks ◽  
Deep Learning ◽  
Automated Generation ◽  
Advantages And Disadvantages ◽  
Automated Learning ◽  
Ice Drift ◽  
Artificial Neural ◽  
The Creation

Goal: the need for systems of automated generation of models of complexly formalized objects is considered. The approach to the creation of such a system based on deep learning is described. Materials and methods: the article describes the architecture of the application of automated learning, based on deep learning, in particular on the basis of the genetic algorithm. Results: the testing of the presented system was carried out on the example of solving the problem of predicting the parameters of ice drift on the Northern Dvina River. Conclusion: the advantages and disadvantages, features of implementation, the scope of the presented system are shown.

Heliport Detection Using Artificial Neural Networks

2020 ◽  
Vol 86 (9) ◽  
pp. 541-546
Author(s):  
Emre Başeski
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Deep Learning ◽  
Learning Approaches ◽  
Remote Sensing Images ◽  
Advantages And Disadvantages ◽  
Learning Techniques ◽  
Critical Technology ◽  
Artificial Neural ◽  
Military Facilities

Automatic image exploitation is a critical technology for quick content analysis of high-resolution remote sensing images. The presence of a heliport on an image usually implies an important facility, such as military facilities. Therefore, detection of heliports can reveal critical information about the content of an image. In this article, two learning-based algorithms are presented that make use of artificial neural networks to detect H-shaped, light-colored heliports. The first algorithm is based on shape analysis of the heliport candidate segments using classical artificial neural networks. The second algorithm uses deep-learning techniques. While deep learning can solve difficult problems successfully, classical-learning approaches can be tuned easily to obtain fast and reasonable results. Therefore, although the main objective of this article is heliport detection, it also compares a deep-learning based approach with a classical learning-based approach and discusses advantages and disadvantages of both techniques.

Prediction of corrections for the Polish time scale UTC(PL) using artificial neural networks

2013 ◽  
Vol 61 (3) ◽  
pp. 589-594 ◽  
Author(s):  
M. Luzar ◽  
Ł. Sobolewski ◽  
W. Miczulski ◽  
J. Korbicz
Keyword(s):  
Neural Networks ◽  
Time Series ◽  
Artificial Neural Networks ◽  
Time Scale ◽  
Theoretical Background ◽  
Group Method ◽  
Prediction Errors ◽  
Advantages And Disadvantages ◽  
Different Types ◽  
Artificial Neural

Abstract In this paper, the effectiveness of using Artificial Neural Networks (ANNs) for predicting the corrections of the Polish time scale UTC(PL) (Universal Coordinated Time) is presented. In particular, prediction results for the different types of neural networks, i.e., the MLP (MultiLayer Perceprton), the RBF (Radial Basis Function) and the GMDH (Group Method of Data Handling) are shown. The main advantages and disadvantages of using such types of neural networks are discussed. The prediction of corrections is performed using two methods: the time series analysis method and the regression method. The input data were prepared suitable for the above mentioned methods, based on two time series, ts1 and ts2. The designation of prediction errors for specified days and the influence of data quantity for the prediction error are considered. The paper consists of five sections. After Introduction, in Sec. 2, the theoretical background for different types of neural networks is presented. Section 3 shows data preparation for the appropriate type of neural network. The experimental results are presented in Sec. 4. Finally, Sec. 5 concludes the paper.

scholarly journals Application of Artificial Neural Networks in Analysis of Time-Variable Optical Reflectance Spectra in Digital Light Projection Spectroscopy

Coatings ◽  
2021 ◽  
Vol 12 (1) ◽  
pp. 37
Author(s):  
Marek Gąsiorowski ◽  
Piotr Szymak ◽  
Leszek Bychto ◽  
Aleksy Patryn
Keyword(s):  
Neural Networks ◽  
Optical Properties ◽  
Artificial Neural Networks ◽  
Mean Squared Error ◽  
Reflectance Spectra ◽  
Training Methods ◽  
Optical Reflectance ◽  
Research Cycle ◽  
Evaluation Module ◽  
Artificial Neural

This article undertakes the subject matter of applying artificial neural networks to analyze optical reflectance spectra of objects exhibiting a change of optical properties in the domain of time. A compact Digital Light Projection NIRscan Nano Evaluation Module spectrometer was used to record spectra. Due to the miniature spectrometer’s size and its simplicity of measurement, it can be used to conduct tests outside of a laboratory. A series of plant-derived objects were used as test subjects with rapidly changing optical properties in the presented research cycle. The application of artificial neural networks made it possible to determine the aging time of plants with a relatively low mean squared error, reaching 0.56 h for the Levenberg–Marquardt backpropagation training method. The results of the other ten training methods for artificial neural networks have been included in the paper.

FORECASTING THE INTENSITY OF SOLAR RADIATION BASED ON ARTIFICIAL NEURAL NETWORKS

2021 ◽  
Vol 43 (2) ◽  
pp. 60-67
Author(s):  
B.I. Basok ◽  
M.P. Novitska ◽  
V.P. Kravchenko
Keyword(s):  
Neural Network ◽  
Experimental Data ◽  
Neural Networks ◽  
Artificial Neural Network ◽  
Artificial Neural Networks ◽  
Solar Radiation ◽  
Data Set ◽  
Advantages And Disadvantages ◽  
Database Evaluation ◽  
Artificial Neural

The paper considers short-term forecasting of the intensity of solar radiation in the city of Odessa based on an artificial neural network. The artificial neural network was trained on the experimental data of the ground weather station (Davis 6162EU), which is installed on the roof of the educational building of the Odessa National Polytechnic University. Modeling, validation, and testing of experimental data were performed using the MATLAB software package, namely Neural Network Toolbox. The Levenberg-Markwatt model is used in this work. The analyzed data set was divided into proportions of 70%, 15%, 15% for neural network training, its validation, and testing, respectively. The results which the trained neural network gave during forecasting within the framework of the database and outside it are given. The deviation between real and forecast data is analyzed. The root-mean-square error on December 26, 2016 was 13.03 W / m2, and on December 27, 2016 - 9.44 W / m2 when forecasting outside the database. Evaluation of the accuracy of an artificial neural network has shown its effectiveness in predicting the intensity of solar radiation. To predict parameters based on artificial neural networks, experimental data that describe a real system are needed. Artificial neural networks, like other approximation methods, have both advantages and disadvantages.

Inverse Kinematic Solutions Using Artificial Neural Networks

2014 ◽  
Vol 534 ◽  
pp. 137-143 ◽  
Author(s):  
Iskandar Petra ◽  
Liyanage C. de Silva
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Inverse Kinematics ◽  
Joint Angle ◽  
Inverse Kinematic ◽  
Task Execution ◽  
Advantages And Disadvantages ◽  
Complex Process ◽  
Inverse Kinematics Problem ◽  
Artificial Neural

Inverse Kinematics solutions are needed for control of robotic manipulators for successful task execution. It is the process of obtaining the required manipulator joint angle values for a given desired end point position and orientation. In general the process of obtaining these joint angle values is a complex process that may require some higher computational power in the hardware. Mainly there are three traditional methods used to solve inverse kinematics problem, namely; geometric methods, algebraic methods and iterative methods. Apart from these traditional techniques researchers have looked into the use of Artificial Neural Networks (ANNs). In this paper we re-visit these non-traditional techniques and compare the advantages and disadvantages of each method.

Sign in / Sign up

Export Citation Format

Share Document