Banca de DEFESA: MARIA FERNANDA CABRAL RIBEIRO

Fault Tolerance in FPGA-based Multilayer Perceptron: Case Study SALVE TODAS

Violence Against Women, Embedded Systems, Artificial Intelligence, Artificial Neural Networks, FPGA, Fault Tolerance.

The concept of fault tolerance can be understood as the ability of a system to maintain its correct operation even after the occurrence of a failure. This area of study emerged in the 1950s, aimed at dealing with shortages in military and aerospace equipment operating in hostile and/or remote environments, and since then it has proven to be a prominent field of study, especially with the popularization of the use of computers and embedded systems.
In this context, this work aims: the application of fault tolerance techniques in an Artificial Neural Network with Multilayer Perceptron (MLP) architecture embedded in an FPGA. The MLP network in question makes up a system aimed at women's safety that aims to identify, through the MLP network, possible risk situations for users. To this end, the system has sensors for vital signs, sudden movements and geolocation that provide information about the user's current situation. Since the MLP Network plays a critical role in identifying risk situations, it is necessary to apply techniques aimed at increasing its reliability, aiming at greater safety for the user. Therefore, this work analyzes the gains and impacts of applying four fault tolerance techniques combined in the embedded MLP. The techniques used include: dealing with the weights and biases of neurons in the network's processing layers; the removal of hidden neurons that are less sensitive to failure; the duplication of hidden neurons that are more sensitive to failures (a technique known as Augmentation); and the Triple Modular Redundancy of the neurons in the input and output layers of the network.