A Standardized Representation of Convolutional Neural Networks for Reliable Deployment of Machine Learning Models in the Manufacturing Industry

Published: August 18, 2019

Author(s)

Max K. Ferguson, Seongwoon Jeong, Kincho H. Law, Anantha Narayanan Narayanan, Svetlana Levitan, Jena Tridivesh, Yung-Tsun T. Lee

Abstract

The use of deep convolutional neural networks is becoming increasingly popular in the engineering and manufacturing sectors. However, managing the distribution of trained models is still a difficult task, partially due to the limitations of standardized methods for neural network representation. This paper seeks to address this issue by proposing a standardized format for convolutional neural networks, based on the Predictive Model Markup Language (PMML). A number of pretrained ImageNet models are converted to the proposed PMML format to demonstrate the flexibility and utility of this format. These models are then fine-tuned to detect casting defects in Xray images. Finally, a high-performance scoring engine is developed to evaluate new input images against models in the proposed format. The utility of the proposed format and scoring engine is demonstrated by bench marking the performance of the defect detection models on a range of different computation platforms. The scoring engine and trained models are made available at https://github.com/maxkferg/python-pmml
Proceedings Title: Proceedings of the ASME 2019 International Design Engineering Technical Conferences & Computers and Information in Engineering Conference IDETC/CIE 2019
Conference Dates: August 18-21, 2019
Conference Location: Anheim, CA
Conference Title: ASME 2019 International Design Engineering Technical Conferences & Computers and Information in Engineering Conference IDETC/CIE 2019
Pub Type: Conferences

Download Paper

Keywords

Machine Learning, Image Processing, Convolutional Neural Networks, PMML
Created August 18, 2019, Updated September 05, 2019