BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//TUC//Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:Europe/Athens
TZNAME:EEST
DTSTART:19700329T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0300
TZNAME:EET
DTSTART:19701025T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CREATED:20230330T074532Z
LAST-MODIFIED:20230330T074532Z
DTSTAMP:20260817T223217Z
UID:1786995137@tuc.gr
SUMMARY:Παρουσίαση Μεταπτυχιακής Εργασίας κ.
  Γεωργίου Αποστολάκη - Σχολή ΗΜΜΥ
LOCATION:
DESCRIPTION:https://www.ece.tuc.gr/el/katalogos-
 ekdiloseon?tx_tucevents2_tuceventsdi
 splay%5Baction%5D=show&tx_tucevents2
 _tuceventsdisplay%5Bcontroller%5D=Ev
 ent&tx_tucevents2_tuceventsdisplay%5
 Bevent%5D=6127&cHash=1b8c312c58cfcfb
 1becb9fa22b77d16f\nΠΟΛΥΤΕΧΝΕΙΟ ΚΡΗΤΗ
 Σ\n Σχολή Ηλεκτρολόγων Μηχανικών και
  Μηχανικών Υπολογιστών\n Πρόγραμμα Μ
 εταπτυχιακών Σπουδών\n ΠΑΡΟΥΣΙΑΣΗ ΜΕ
 ΤΑΠΤΥΧΙΑΚΗΣ ΕΡΓΑΣΙΑΣ\n ΓΕΩΡΓΙΟΥ ΑΠΟΣ
 ΤΟΛΑΚΗ\n με θέμα\n Τεχνικές Συμπερασ
 μού σε Δίκτυα Αισθητήρων Χαμηλού Κόσ
 τους\n Inference Techniques in Low-C
 ost Sensor Networks\n Εξεταστική Επι
 τροπή\n Καθηγητής Άγγελος Μπλέτσας (
 επιβλέπων)\n Καθηγητής Μιχαήλ Γ. Λαγ
 ουδάκης\n Καθηγητής Αντώνιος Δεληγια
 ννάκης\n Abstract\n Distributed exec
 ution of algorithms across resource-
 constrained terminals has become inc
 reasingly popular, especially when f
 ault tolerance is required. Asynchro
 nous operation is brought to light i
 n such scenarios, and in particular,
  probabilistic asynchronous operatio
 n, which models the failure probabil
 ity of each terminal. The focus of t
 his work is on the affine update mod
 el, which is applicable to a wide ra
 nge of distributed inference algorit
 hms. Applications include estimation
  of the average, solving linear syst
 ems, linear minimum mean square erro
 r estimation and spectral clustering
 , presented in detail in this thesis
 . Multimodal inference is also inves
 tigated, where two or more alternate
  sources of data are exploited for i
 ncreased prediction accuracy. In tha
 t context, two variations of linear 
 regression are presented, with unifo
 rm or Gaussian prior, which are equi
 valent to iterative affine updates. 
 Furthermore, this work offers an asy
 mptotic analysis for the arithmetic 
 mean of the state vector, across a f
 inite number of experiments, for the
  discovery of fixed points. It is sh
 own that there are cases where the a
 rithmetic mean behaves differently t
 han the expected mean, and a suffici
 ent condition is provided for conver
 gence of the arithmetic mean to a fi
 xed point. The lack of necessity for
  this condition is explained and sub
 cases where the arithmetic mean conv
 erges, diverges, or has unpredictabl
 e behaviour are highlighted. Additio
 nally, cases where the individual it
 erations never converge but their ar
 ithmetic mean does and offers fixed 
 point are offered. Simulations corro
 borate the theoretical findings for 
 various affine model setups. Finally
 , implementation details of a distri
 buted low-cost sensor network are pr
 esented; the low-cost sensor network
  estimates the arithmetic mean of te
 mperature across the network, by exe
 cuting average consensus. The implem
 entation is divided into hardware, n
 etwork and software layers, and each
  layer is presented separately.\n Me
 eting ID: 917 3783 0978\n Password: 
 473502\n
STATUS:CONFIRMED
ORGANIZER;RSVP=FALSE;CN=TUC;CUTYPE=TUC:mailto:webmaster@tuc.gr
DTSTART:20230405T150000
DTEND:20230405T160000
TRANSP:OPAQUE
CLASS:DEFAULT
END:VEVENT
END:VCALENDAR