BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//labri.fr//NONSGML kigkonsult.se iCalcreator 2.41.92//
CALSCALE:GREGORIAN
METHOD:PUBLISH
UID:05540db6-5823-45f3-9118-0f6cbb36b646
X-WR-CALNAME:[M2F] Gabriel Bathie - 'LTL Learning meets Boolean Set Cover: 
 enumerative algorithms for program synthesis'
X-WR-TIMEZONE:Europe/Paris
BEGIN:VTIMEZONE
TZID:Europe/Paris
TZUNTIL:20271031T010000Z
BEGIN:STANDARD
TZNAME:CET
DTSTART:20251026T030000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
RDATE:20261025T030000
END:STANDARD
BEGIN:DAYLIGHT
TZNAME:CEST
DTSTART:20250330T020000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
RDATE:20260329T020000
RDATE:20270328T020000
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:05540db6-5823-45f3-9118-0f6cbb36b646
DTSTAMP:20260511T094646Z
CLASS:PUBLIC
DESCRIPTION:In automated learning problems\, the task is to find a model th
 at maps given inputs to their corresponding outputs as accurately as possi
 ble. Over the past 30 years\, machine learning and deep learning have achi
 eved tremendous success in solving this type of problem. However\, while t
 he resulting models can be used to make predictions\, they offer limited i
 nterpretability\, i.e. they provide little insight into *how* they solve t
 he problem. For instance\, if we train a recurrent neural network to predi
 ct whether a sequence of events will lead to a crash\, the model cannot de
 scribe the sequences of events that lead to crashes and provide solutions 
 to remediate the issue. Program synthesis is a framework for solving learn
 ing problems with models that are programs in a domain-specific language\,
  which allows creating interpretable models in the domain of the problem.
 \n\nIn this talk\, I will present an enumerative approach for learning Lin
 ear Temporal Logic (LTL) formulas from data. I will begin with an introduc
 tion to program synthesis\, using examples from program de-obfuscation and
  anomaly explanation. In the second part\, I will present the main techniq
 ues used in our algorithm: observational equivalence and domination\, two 
 pruning techniques used to reduce the search space\, and a connection betw
 een LTL Learning and the Boolean Set Cover problem. Finally\, I will discu
 ss the engineering choices we made when implementing this algorithm in Bol
 t\, an open-source tool available at https://github.com/SynthesisLab/Bolt.
 \n\nThis talk is based on joint work with Nathanaël Fijalkow\, Théo Matric
 on\, Baptiste Mouillon and Pierre Vandenhove. \n\n\nImport automatique dep
 uis https://framagenda.org/remote.php/dav/public-calendars/D3yqF7nwys9amfy
 Y?export par sync_icals_to_drupal.py pour M2F
DTSTART;TZID=Europe/Paris:20260324T140000
DTEND;TZID=Europe/Paris:20260324T150000
LOCATION:LaBRI
SEQUENCE:0
SUMMARY:[M2F] Gabriel Bathie - 'LTL Learning meets Boolean Set Cover: enume
 rative algorithms for program synthesis'
TRANSP:OPAQUE
END:VEVENT
END:VCALENDAR
