A high performance cognitive framework (SIVA - self intelligent versatile and adaptive) for heterogenous architecture in IOT environment

The advent of the IOT has brought automation to our footsteps. But the darker side of this technology is to implement machine learning for intelligent detection. Several machine learning algorithms like artificial neural networks, support vector machines, deep learning are applied for bringing the cognitive aspects in internet of things. These algorithms find their applications in face, emotion recognition's, etc., on the hardware. But there is a need for developing low power, high accurate, intelligent machine learning framework for embedded architectures for dynamic inputs in health care solutions. Hence we propose a framework named self intelligent versatile and adaptive (SIVA) for dynamic inputs in IOT-based healthcare solutions. This framework is based on neural network and cognitive rule sets for self-learning and adaptability. The proposed learning algorithm works on self-adaptive principles which make the framework suitable for wearable devices with dynamic inputs. This framework has been evaluated for different biomedical sensors and embedded heterogeneous architectures. Various performance parameters viz. recognition rate, accuracy, execution time and energy are measured and analysed. The results indicate that the framework not only have superiority on complexity, but also have low power consumption over existing algorithms.