BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20190719T085743Z
LOCATION:HG D 3.2
DTSTART;TZID=Europe/Stockholm:20190612T130000
DTEND;TZID=Europe/Stockholm:20190612T133000
UID:submissions.pasc-conference.org_PASC19_sess124_msa179@linklings.com
SUMMARY:Will Artificial Intelligence Replace Computational Economists Any 
 Time Soon?
DESCRIPTION:Minisymposium\nEmerging Application Domains\n\nWill Artificial
  Intelligence Replace Computational Economists Any Time Soon?\n\nMaliar, M
 aliar, Winant\n\nIn this paper, we ask whether a generic artificial intell
 igence-style (AI) algorithm is capable of replacing computational economis
 ts so that there is no need to write model-specific codes. We analyze thre
 e examples: a stylized consumption-saving problem with borrowing constrain
 ts, large-scale OLG model and a new Keynesian model with ZLB on the nomina
 l interest rate. We show how to cast such models into the form suitable fo
 r AI, and we solve it with standard AI software, namely, Google's TensorFl
 ow library. Our approximation builds on multilayer neural networks, which 
 we train by using deep learning (DL) techniques. We find the neural networ
 ks to be flexible, scalable and tractable in large scale applications. We 
 also experiment with several other types of approximating functions includ
 ing polynomials and splines, and we document important tradeoffs between t
 he accuracy and cost.  
END:VEVENT
END:VCALENDAR