Univerza na Primorskem Fakulteta za matematiko, naravoslovje in informacijske tehnologije
SI | EN

sreda, 2. junij 2021 Niki Hrovatin: In-Network Convolution in grid shaped Sensor Networks

V ponedeljek, 31. maja 2021, bo ob 16.00 uri prek spletnih orodij na daljavo izvedeno predavanje v okviru PONEDELJKOVEGA SEMINARJA RAČUNALNIŠTVA IN INFORMATIKE Oddelkov za Informacijske znanosti in tehnologije UP FAMNIT in UP IAM.

ČAS/PROSTOR: 31. maj 2021 ob 16.00 na daljavo

-----------------------------------------
PREDAVATELJ: Niki HROVATIN
-----------------------------------------

Niki Hrovatin is a teaching assistant and PhD student at the University of Primorska, Department of Information Sciences and Technologies, and an Assistant Researcher at the InnoRenew CoE. His current research interests include Wireless Sensor Networks, Security and Privacy, Machine Learning, and Blockchain.

------------------------------------------------------------------------------------------
NASLOV: In-Network Convolution in grid shaped Sensor Networks
------------------------------------------------------------------------------------------

POVZETEK: 

A Sensor Network consists of spatially distributed nodes deployed in a dynamic environment for specific monitoring purposes. We present the recent development of our use-case sensor network, a smart floor designed to detect falls. The smart floor is a grid-shaped sensor network in which each sensor node is sensing the local force applied on the floor. A Convolutional Neural Network (CNN) is used to recognize if the activity occurring over the smart floor is a person that fell on the floor or just activities of daily living like walking, moving objects, etc. Furthermore, we discuss a technique applicable only to grid-shaped sensor networks for distributed processing of convolutional layers on sensor nodes.

---------------------------------------------------------------------------------------------------

Predavanje bo potekalo v angleškem jeziku prek spletnega orodja Zoom.
Do predavanja dostopate tako, da se povežete prek sledeče povezave:

https://upr-si.zoom.us/j/297328207

Vabljeni!