Fabric defect detection have importance in terms of sectoral quality. Automatic systems are developed on the defect detection, in this regard many methods are tried to obtain high precision with image processing studies. In this study, deep learning which distinguishes with multi-layer architectures and reveals high achievement is applied to fabric defect detection. Autoencoder -a deep learning algorithm- that aimed to represent input data via compression or decompression is tried to detect defect of fabrics and it gains acceptable success. The vital goal of this study is to increase achievement of feature extraction by tuning up the autoencoder's input value and hyper parameters.
1 September 2020 in Journal of Molecular Graphics and Modelling
Improvements in technology have triggered the production of big data. Within this scope, enormous amounts of biological data have been generated. A number of analysis methods have been developed to access the information contained in biological data. DNA sequence analysis has drawn particular attention in recent years. As an alternative to alignment-based sequence comparison methods that have high computational costs, alignment-free comparison methods have emerged. These m ethods c an c alculate sequence s imilarity by applying different dimensions of numerical characterizations. In this paper, we propose a novel alignment-free DNA sequence analysis method based on a feature extraction strategy. The method utilizes numerical characterization and is implemented by calculating mean distance of the transitions, mean distance of the nucleotide duplications, and the base frequencies. The method then measures the similarity between 7-dimensional vectors that are obtained through feature extraction. Using this approach, we conducted a sequence similarity analysis of two different DNA sequence datasets of different lengths to demonstrate the effectiveness of the method. The proposed method shows that a simple and successful feature vector can be obtained when DNA sequences having many properties are used in combination with appropriate and effective descriptors. With this strategy, reasonable results were obtained with a low computational cost.