Right now we are living Artificial Intelligence (AI) boom bubble, where you can look the big progress that has been cooking in the labs for more than 30 years, some of that years the progress were stuck waiting for more powerful computers and algorithms’ improve, but now every day we see artificial intelligence systems that process huge amounts of data and make projections or take decision-based in huge computer clusters with hundreds or thousands of CPUs, GPUs and terabytes of memory and HDD storage, and in this case the AI is at the hand of Big Data because usually in machine learning and deep learning is needed a lot of information, usually tagged, to train the AI algorithms. But at the other end we can have machine learning and deep learning in embedded systems just based on the data recorded in the local environment, this data is known as “Tiny Data”.
In this embedded environment, the AI system has more challenges that his bigger brothers, because the computational resources like CPU, memory, storage and power are more limited and also it is part of a real-time system that needs to meet the deadlines to accomplish the main tasks expected by the users; also, the AI system need to deal with on-device data collection, which by its nature, will be less plentiful, unlabeled, but at least it is more targeted to the embedded application.
For example, at BolivarTech we just finished one costumer’s project where we provide the AI algorithms to be incorporated at one industrial water heater, because the costumer want to improve the efficiency by allowing the water heater learn the water temperature requirements and behaviors in it particular environment; based on this requirements the water heater need to collect data like water flow, ambient temperature, target temperature, time to reach the target temperature based in the incoming water temperature, etc. and based in this data collected predict the duty cycle needed in order to reduce the consumed energy to keep the water at the target temperature.
This is one example where the AI and the Tiny Data work together in order to improve one embedded system, providing to it a real smart behavior, for this in your next smart project contact us at BolivarTech about your artificial intelligence needs.
Julian Bolivar-Galeno is an Information and Communications Technologies (ICT) Architect whose expertise is in telecommunications, security and embedded systems. He works in BolivarTech focused on decision making, leadership, management and execution of projects oriented to develop strong security algorithms, Artificial Intelligence (AI) research and its applicability to smart solutions at mobile and embedded technologies, always producing resilient and innovative applications.
Embedded AI and Tiny Data
