logo SBA

ETD

Archivio digitale delle tesi discusse presso l’Università di Pisa

Tesi etd-10162021-103039


Tipo di tesi
Tesi di laurea magistrale
Autore
PAMPALONI, SIMONE
URN
etd-10162021-103039
Titolo
Designing of Distributed Learning Infrastructure with Latency-Energy Tradeoffs
Dipartimento
INGEGNERIA DELL'INFORMAZIONE
Corso di studi
COMPUTER ENGINEERING
Relatori
relatore Prof. Tonellotto, Nicola
relatore Prof. Vallati, Carlo
Parole chiave
  • distributed learning
  • energy-latency tradeoffs
  • federated learning
  • flower
  • jetson
Data inizio appello
19/11/2021
Consultabilità
Non consultabile
Data di rilascio
19/11/2024
Riassunto
Artificial intelligence (AI) and machine learning (ML) have become of crucial importance in many scientific and industrial fields, thanks to the ability to extract information, make predictions and identify patterns on data. For the creation of increasingly accurate predictive models, these technologies are based on the collection and control of large amounts of data within controlled systems. Federated Learning (FL) are new settings that exploit the computational capabilities and local data of a set of devices coordinated by a central server for the creation of a shared global predictive model, without the need for collect data. The objective of this thesis is to implement a distributed infrastructure exploiting Flower, a framework that facilitates the implementation of distributed and federated systems, using two versions of NVIDIA's Jetson as devices, studying and analyzing the characteristics of this infrastructure, and finally design a model that allows to analyze the tradeoffs between system energy consumed and system latency, by varying the amount of data that each device makes available for the development of the global predictive model ML.
File