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:20220812T074334Z
LOCATION:Sydney Room
DTSTART;TZID=Europe/Stockholm:20220627T140000
DTEND;TZID=Europe/Stockholm:20220627T143000
UID:submissions.pasc-conference.org_PASC22_sess117_msa275@linklings.com
SUMMARY:Infero: An API for Machine Learning Inference in Operations
DESCRIPTION:Minisymposium\n\nInfero: An API for Machine Learning Inference
  in Operations\n\nBonanni, Chantry, Hawkes, Dueben, Quintino...\n\nMachine
  Learning models are often assembled and trained within a python environme
 nt. In most cases, inference is also run in the same environment, and no a
 dditional programming effort is required. Nevertheless, there are cases wh
 en the trained machine learning model must be used for inference within (o
 r alongside) a pre-existing application. In such cases an interface must b
 e setup. This interface can be implemented in different ways and many solu
 tions are being explored and tested by different research groups. In this 
 work we present a library called "Infero" that allows an HPC application t
 o directly call a pre-trained ML model for inference. <em>Infero</em> prov
 ides an interface accessible from multiple programming languages (Python, 
 C, C++ and Fortran), allowing the application to pass in n-dimensional Ten
 sors with input data, and get back the m-dimensional output Tensors. Inter
 nally, <em>Infero</em> has a back-end architecture where various types of 
 runtime engines can be used to perform the inference step via a common int
 erface (back-ends currently supported: TensorFlow Lite, TensorFlow C-API, 
 ONNX runtime and TensorRT). This presentation covers the key elements of t
 he development of <em>Infero</em> and some preliminary use-cases and resul
 ts.\n\nDomain: Climate, Weather and Earth Sciences, Computer Science and A
 pplied Mathematics
END:VEVENT
END:VCALENDAR