A TANGO Framework Technology that Supports Neural Network Deployment Considering the Kubernetes Environment

Artificial intelligence service development requires many procedures, including neural network creation, neural network optimization, application template code, and deployment. The kubernetes environment is very useful for deploying and managing neural networks. However, these processes require very specialized knowledge and skills. Therefore, tools that can automate these procedures more easily are required. This paper presents a neural network automatic generation and deployment framework technology that can automatically generate and deploy neural networks. In particular, the deployment implemented a neural network service by deploying a neural network and application template in a kubernetes environment. Kubernetes-based deployment is very useful for cloud-based deployments, such as efficient resource distribution, simplified deployment, and service recovery. The proposed technique provides a method to easily and quickly build a neural network application system by deploying neural networks and template codes optimized for kubernetes-based target systems.