Cross-Stitch Multi Task Feature Learning For Resource Allocation in IOT

Abstract

Growth in network technologies has led the world towards the era of disruptive technologies like IOT, Machine Learning etc. With growing technology like IOT, the generation of data and devices are increasing tremendously. So resource allocation problem in such environment has become a major research area. There are many existing machine learning and deep learning solutions for resource allocation problem but these models are trained for specific application and to solve a specific problem. So, small changes in nature of problem will require restructuring of architecture and retraining of model on timely manner which is cumbersome task for real time systems. I proposed a multi-task learning based approach to handle resource allocation problem. Deep learning multi-task will help in generalization of resource allocation problem and to tackle rare situation related to this problem. Multi-task learning will help to generalize machine learning solutions for multiple tasks by sharing some of the information between them.