BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:America/Denver
BEGIN:DAYLIGHT
DTSTART:20250309T030000
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3
TZNAME:MDT
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20251102T010000
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11
TZNAME:MST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20251026T031547Z
UID:32BC025C-1D57-47B7-B562-01F00EB64A89
DTSTART;TZID=America/Denver:20251023T170000
DTEND;TZID=America/Denver:20251023T190000
DESCRIPTION:Abstract: Ensuring that AI systems do what we\, as humans\, act
 ually want them to do\, is one of the biggest open research challenges in 
 AI alignment and safety. Dr. Brown's research seeks to directly address th
 is challenge by enabling AI systems to interact with humans to learn align
 ed and robust behaviors. The way robots and other AI systems behave is oft
 en the result of optimizing a reward function. However\, manually designin
 g good reward functions is highly challenging and error prone\, even for d
 omain experts. Although reward functions for complex tasks are difficult t
 o manually specify\, human feedback in the form of demonstrations or prefe
 rences is often much easier to obtain. However\, human data is often diffi
 cult to interpret due to ambiguity and noise. Thus\, it is critical that A
 I systems take into account uncertainty over the human's true intent. Dr. 
 Brown's talk will give an overview of my lab's progress along the followin
 g fundamental research areas: (1) efficiently maintaining uncertainty over
  human intent\, (2) directly optimizing behavior to be robust to uncertain
 ty\, and (3) actively querying for additional human input to reduce uncert
 ainty.\n\nSpeaker(s): \, Dr. Brown\n\nVirtual: https://events.vtools.ieee.
 org/m/505194
LOCATION:Virtual: https://events.vtools.ieee.org/m/505194
ORGANIZER:s.mehalingam.us@ieee.org
SEQUENCE:17
SUMMARY:Toward Robust\, Interactive\, and Human-Aligned AI Systems
URL;VALUE=URI:https://events.vtools.ieee.org/m/505194
X-ALT-DESC:Description: <br /><div style="font-family: Aptos\,Aptos_Embedde
 dFont\,Aptos_MSFontService\,Calibri\,Helvetica\,sans-serif\; font-size: 12
 pt\; color: rgb(0\,0\,0)\;"><strong>Abstract: </strong>Ensuring that AI sy
 stems do what we\, as humans\, actually want them to do\, is one of the bi
 ggest open research challenges in AI alignment and safety. Dr. Brown's res
 earch seeks to directly address this challenge by enabling AI systems to i
 nteract with humans to learn aligned and robust behaviors. The way robots 
 and other AI systems behave is often the result of optimizing a reward fun
 ction. However\, manually designing good reward functions is highly challe
 nging and error prone\, even for domain experts. Although reward functions
  for complex tasks are difficult to manually specify\, human feedback in t
 he form of demonstrations or preferences is often much easier to obtain. H
 owever\, human data is often difficult to interpret due to ambiguity and n
 oise. Thus\, it is critical that AI systems take into account uncertainty 
 over the human's true intent. Dr. Brown's talk will give an overview of my
  lab's progress along the following fundamental research areas: (1) effici
 ently maintaining uncertainty over human intent\, (2) directly optimizing 
 behavior to be robust to uncertainty\, and (3) actively querying for addit
 ional human input to reduce uncertainty.</div>\n<div style="font-family: A
 ptos\,Aptos_EmbeddedFont\,Aptos_MSFontService\,Calibri\,Helvetica\,sans-se
 rif\; font-size: 12pt\; color: rgb(0\,0\,0)\;">&nbsp\;</div>
END:VEVENT
END:VCALENDAR