From:: tar@medg.lcs.mit.edu (Thomas A. Russ)
Date: Thu, 9 May 91 15:03:23 EDT
To: gsl@ai.mit.edu
Subject: Robust Methods for Graduate Digestion

TIME AND LOCATION: 12:10pm, Friday May 10, Room NE43 8th Floor Playroom

TITLE: Robust Methods for Graduate Digestion

SPEAKER: Cal Q. Later, AI Laboratory, University of Munchagain

An overview of work in applying robust methods to problems in graduate
digestion will be presented.  The talk will stress  why robust methods
are necessary and  appropriate   to  problems  in digestion.    Robust
methods reject  starvation, preserve good  mental health,  and provide
self-service  and self-feeding  capabilities.   Examples from  topping
choice theory,   topology and serving  fitting  in  gustatory imagery,
dynamic  digestion for  mobile  students, and  reliable  navigation to
buffets from unreliable and cryptic announcements will be presented.

The two stage algorithm for  topping  choice can grid sparse data such
as small amounts of pepperone, reject topping and pie edge-overlap due
to  correspondence  mismatches, and fit  a weighted product-cost cubic
spline trade-off function to  the data while preserving small business
profitability.  Results with real and creative data will be presented.

Among  mobile students, our  algorithm  for cheese fusion  can combine
philosphical and ethnic taste disparities  formed from multiple inputs
and determine a satisfying menu without assumptions about the scene or
political  views of  the  students.   The student can  determine   the
structure   of  his/her  surroundings   and correct  emotion  readings
obtained from dead reckoning using  egomotion.  Results from an indoor
scene will be shown.  We present algorithms to determine position when
nearly half of the sightings of leftovers are false.


Hosts:  David Jacobs, Todd Cass, Tao Alter
Coordinator: Karen Sarachik
Confusion Sower: Tom Russ