Algebraic Information Theory (Monday, Oct 24, 14.30 @IDSIA Meeting Room)
by Announcements of talks＠IDSIA
Talk Announcement
Monday, October 24, 2016, 14:30
IDSIA Meeting Room
Prof. Marc Pouly, Lucerne University of Applied Sciences and Arts
Algebraic Information Theory.
A talk with definitely more open questions than answers.
Abstract. Quantitative information theories, such as Hartley’s
relational information theory or Shannon’s statistical information
theory, have beyond doubt been a success story. However, these
theories are based on the fundamental assumption that information can
be measured and expressed in bits. Alternatively, the valuation
algebra framework proposed by Shenoy and Kohlas, which is known to
provide the formal requirements for the application of variable
elimination and other inference schemes, equipped with an additional
axiom allows for the definition of a qualitative information theory
that in a generic way establishes an order of information without
quantifying individual information pieces. As with many other
information theories it is to a large extend an open question how
these theories relate to each other and whether they are consistent,
but recent investigations indicate that specific formalisms of
imprecise probabilities could provide the missing link for such a
deeper understanding.
About the speaker. Marc Pouly studied computer science and mathematics
at the University of Fribourg in Switzerland. After his PhD in
artificial intelligence and information theory, Marc joined the Cork
Constraint Computation Center in Ireland, where he worked on discrete
optimization problems and constraint programming, and later the
Interdisciplinary Center for Security, Reliability and Trust at the
University of Luxembourg, where he investigated the application of
different information theories in the context of information security.
Since 2011 Marc Pouly is professor at the Lucerne University of
Applied Sciences and Arts in Switzerland and since recently head of
the computer science master¹s program. He has been active in teaching
artificial intelligence and information security and pursues various
research projects in the context of industrial optimization, machine
learning and security.
8 years, 2 months
- 1
- 0
08.11 1130 M Dell'Amico: A batching moves iterated local search algorithm for the bin packing problem with generalized precedence constraints
by Announcements of talks＠IDSIA
08 Nov 1130-1200
IDSIA meeting room, Galleria 1, Manno
Prof Mauro Dell'Amico
University of Modena and Reggio Emilia, Italy
Title: A batching moves iterated local search algorithm
for the bin packing problem with generalized precedence
constraints
Abstract: We propose a generalization of the bin packing
problem, obtained by adding precedences between items that
can assume heterogeneous non-negative integer values. Such
generalization also models the well-known Simple Assembly
Line Balancing Problem of type I. To solve the problem we
introduce a simple and effective iterated local search
algorithm that integrates two constructive procedures and
four neighborhood structures to guide the search to local
optimal solutions. Moreover, we apply some preprocessing
procedures and adapt classical lower bounds from the
literature. Extensive computational experiments on
benchmark instances suggest that the developed algorithm
is able to generate good quality solutions in a reasonable
computational time.
(joint work with R. Kramer and M. Iori)
8 years, 2 months
- 1
- 0
- ← Newer
- 1
- Older →