Regresar

Behavioral Signal Processing with Machine Learning Based on FPGA

Abstract:

This paper focuses on analyzing health problems derived from a sedentary lifestyle. Studies seeking to improve physical activity have shown that a good incentive to increase physical activity requires social feedback, allowing the subject to keep motivated and competitive, along with a feedback of number of steps at the end of the day. This work describes the training and implementation of a neural network as an artificial intelligence model to predict the behavior of an individual, taking advantage of the flexibility provided by Field Programmable Gate Arrays (FPGAs). We propose the design of an edge computing system, analyzing the efficiency on power, area and computational performance. The results are presented through a display, making a comparison of the predicted and expected steps.