Deep learning neural classification for structure-property modelling with engineering alloys. (2022)

• Record Type: Journal Article
• Authors: Siva Raja, P.M.; Sumithra, R.P.; Vidhya, S.; Ramanan, K.
• Is Part Of: Materials today. Volume 62:Part 12(2022)
• Page Start: 6844
• Page End: 6847
• Publication Date: 2022
• Subjects: Integrated Computational Materials Engineering -- Tversky similarity function -- Deep learning classification -- Structure–property linkages[maximum of 6]; Materials science -- Congresses -- Periodicals; 620.1
• DOI: 10.1016/j.matpr.2022.05.051
• Languages: English
• ISSNs: 2214-7853

Abstract

Abstract: Integrated Computational Materials Engineering (ICME) is the method used for performing material discovery and design. Computational techniques presented a new deep learning classification method to screen the candidate material designs. The materials are adapted. In ICME process–structure–property workflows, ambiguity was ubiquitous. A Piecewise Regressive Tversky Similarity based Deep Learning Neural Classification (PRTS-DLNC) Method is introduced for minimizing as well as transmit uncertainties for robust uncertainties. PRTS-DLNC Method has number of compound data are considered as input. An input compound information were given to hidden layer 1. In that layer, piecewise regression is employed for performing the compound data analysis with structure–property linkages. After that, the regression coefficient value is sent to the hidden layer 2. In that layer, tversky similarity function is used to identify the similarity between the regression coefficient value of training compound data and threshold value. Ttversky similarity value varies from 0 to 1 and the results are transmitted to the output layer. By this way, PRTS-DLNC Method improves the performance of structure–property linkages. The computational cost of proposed PRTS-DLNC Method is higher than conventional uncertainty quantification.