BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:MechE Colloquium: Nothing is for Granted: Making Wise Decisions th
 rough Real-time Intelligence
DTSTART:20211102T121500
DTEND:20211102T131500
DTSTAMP:20260609T100702Z
UID:d0c779190267c8ec89d8c3daee5e34521a2736eb6e7f95c663fcbc56
CATEGORIES:Conferences - Seminars
DESCRIPTION:Prof. Anastasia Ailamaki\, Data-Intensive Applications and Sys
 tems Laboratory (DIAS)\, School of Computer and Communication Sciences (IC
 )\, Institute of Computer and Communication Sciences (IINFCOM)\, EPFL\nAbs
 tract: \nIn today's ever-growing demand for fast\, data-driven decisions\
 , variable data formats\, dynamic application requirements\, and the intro
 duction of new hardware platforms severely undermines efficiency of data p
 ipelines. The variety in data formats and workloads forces data pipelines
  to be manually split across a variety of task-specialized systems and co
 mbined through expensive transformation and orchestration processes\, or t
 o adapt both the data and the workloads to match the requirements of a si
 ngle-system\, sacrificing expressiveness and structural information. Fur
 thermore\, the ever-increasing hardware heterogeneity causes task-based sp
 ecialization of the tools to specific hardware such as CPUs or GPUs\, for
 cing a trade-off: designing optimized hardware often means wasting accele
 rator-level parallelism opportunities or tolerating slow and unnecessary c
 ommunication between devices. Typically\, data processing is adapted to 
 the pre-determined data processing system architecture\, losing valuable 
 information in the translation.\n\nReal-time intelligence means to abolish
  all prior preparation and instead make all decisions during execution\, w
 hen all relevant information is available for optimal utilisation of reso
 urces. The system learns and extracts information about the requests\, ins
 tead of depending on pre-determined workload expectations. I will show ho
 w designing top-down the system architecture to allow a data- and workload
 -driven just-in-time specialization enables fast query execution over unpr
 epared\, potentially dirty data without time consuming preparation\, as w
 ell as efficient orchestration and utilization of heterogeneous hardware 
 devices.\n\nBio:\nAnastasia Ailamaki is a Professor of Computer and Commun
 ication Sciences at the École Polytechnique Fédérale de Lausanne (EPFL
 ) in Switzerland and the co-founder of RAW Labs SA\, a Swiss company deve
 loping real-time analytics infrastructures for heterogeneous big data from
  multiple sources. She earned a Ph.D. in Computer Science from the Univer
 sity of Wisconsin-Madison in 2000. She received the 2019 ACM SIGMOD Edgar
  F. Codd Innovations and the 2020 VLDB Women in Database Research Award. S
 he is also the recipient of an ERC Consolidator Award (2013)\, the Finmec
 canica endowed chair from the Computer Science Department at Carnegie Mel
 lon (2007)\, a European Young Investigator Award from the European Scienc
 e Foundation (2007)\, an Alfred P. Sloan Research Fellowship (2005)\, an N
 SF CAREER award (2002)\, and ten best-paper awards in database\, storage\
 , and computer architecture conferences. She is an ACM fellow\, an IEEE f
 ellow\, the Laureate for the 2018 Nemitsas Prize in Computer Science\, and
  an elected member of the Swiss\, the Belgian\, the Greek\, and the Cypri
 ot National Research Councils. She is a member of the Academia Europaea a
 nd of the World Economic Forum Expert Network.
LOCATION:BM 5202 https://plan.epfl.ch/?room==BM%205202 https://epfl.zoom.u
 s/j/65093257313
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR