Constructing a method for the conversion of numerical data in order to train the deep neural networks
Journal Title: Восточно-Европейский журнал передовых технологий - Year 2018, Vol 5, Issue 4
Abstract
<p>This paper analyzes known types of deep neural networks, the methods of their supervised training, training the networks to suppress noise, as well as methods for encoding data using images. It has been shown that deep neural networks are suitable in order to effectively solve classification problems, in particular for medical and technical diagnosing. Among the deep networks, the convolutional neural networks are promising because of their simple structure and application of common weights, which makes it possible for a network to separate similar features in different parts of images. Training a convolutional network may prove insufficient for some diagnosing tasks, which is why it is advisable to consider modifications to the training method using data encoding and training to suppress noise in order to obtain a better result.</p><p>We have proposed a method for training a convolutional neural network using numerical data converted to bitmap images, which would improve the accuracy of a network when solving the problems on classification and which would make it possible to apply the convolutional neural networks and their advantages in image processing by using tabular data as input. In addition, the proposed method requires no additional changes to the structure of the network.</p><p>The method consists of four stages – the normalization using a method of min-max, conversion of data into two-dimensional images applying the float or thermometric encoding methods, the generation of additional images with the distortion of input data, and the preliminary training of a deep network.</p>The constructed method was implemented in software and investigated when solving a number of practical tasks. Results of solving the practical tasks on technical and medical diagnosing have shown the effectiveness of the method at small numbers of the resulting classes and training instances. The method could prove useful when diagnosing a defect at the early stages of its manifestation when the volume of training data is limited
Authors and Affiliations
Mykhailo Pryshliak, Sergey Subbotin, Andrii Oliinyk
Keywords
Related Articles
- EP ID EP528171
- DOI 10.15587/1729-4061.2018.145586
- Views 69
- Downloads 0
Formalization of selection of contract-organizational project delivery strategy
<p>The formalized system to select a strategy for the implementation of a capital construction project was developed. This is important because an error in making a strategic decision (in an intuitive rather than formali...
Development of a sol-gel technique for obtaining sintering activators for engobe coatings
<p>We have developed a technology for manufacturing the glass ligaments-activators to intensify the sintering of ceramic materials, specifically engobe coatings. The comparative analysis has been performed into the glass...
The development of methods for determining vibration stochastic fields of technological complexes
<p>The force effects occurring in technological complexes have been studied on the basis of the analysis of technical diagnostics system. Due to the distinction between deterministic and random force effects, there have...
Development of technology for the production of semifinished products with an emulsion structure based on the decalcified dairy raw materials
<p>It was determined that in a condition of meeting requirements of current normative documents, physicochemical and technological properties of the cottage cheese differ from each other which affects technological param...
Development of a method for estimating the resistance of fibers and threads to a sliding bend based on energy consumption for external and internal friction
<p>We present materials for constructing an instrumental method for assessing resistance of threads to the sliding bend relative to cylindrical surfaces in order to solve tasks on control and prediction of conditions for...