|
|
GENERAL |
|
|
|
|
RESEARCH AREAS |
|
|
|
|
MISCELLANEOUS |
|
|
|
|
Adaptive Robust Computing
Overview
The basis and strength of the Adaptive Robust Computing group rests
on our competence in mathematics.
Stochastic Pattern Computation has been a major focus for many years,
exploring methods from mathematical statistics.
Today the main instruments are algebra (data security protols) and
analysis (optimal control).
Current research areas:
- Protocols for secure multiparty computations
- Methods for optimal control
Projects
Researchers
Publications
2000
Arnborg, S. and G. Sjödin. 2000.
"Bayes Rules in Finite Models." In Proc. of the European Conference on Artificial Intelligence, Berlin, 2000, 571-575.
ISBN 1 58603 013 2
ftp://ftp.nada.kth.se/pub/documents/Theory/Stefan-Arnborg/fobc1.ps
Arnborg, S. and G. Sjödin. 2000.
"On the Foundations of Bayesianism." Presented at the Twentieth International Workshop on Bayesian Inference and Maximum
Entropy Methods in Science and Engineering, 2000
ftp://ftp.nada.kth.se/pub/documents/Theory/Stefan-Arnborg/fobm1.ps
Kanerva, P., J. Kristoferson, A. Holst. 2000.
"Random indexing of text samples for Latent Semantic Analysis."
In Proc. 22nd Annual Conference of the Cognitive Science Society, U. Pennsylvania, Aug. 2000, edited by L.R. Gleitman and
A.K. Josh, p. 1036. Mahwah, New Jersey: Erlbaum, 2000.
Kanerva, P. 2000.
"Large patterns make great symbols: An example of learning from example."
In Hybrid Neural Systems edited by S. Wermter and R. Sun, 194-203.
Heidelberg: Springer, 2000.
2001
Arnborg, S. and G. Sjödin.
On the Foundations of Bayesianism.
In the Twentieth International Workshop on Bayesian Inference and Maximum Entropy Methods in Science and Engineering,
American Institute of Physics, 2001, Ali Mohammad-Djarafi, ed., pages 61-71, ISBN 0-7354-0004-0.
Abstract: We discuss precise assumptions entailing Bayesianism in the line of investigations started by Cox, and relate
them to a recent critique by Halpern. We show that every finite model which cannot be rescaled to probability violates
a natural and simple refinability principle. A new condition, separability, was found sufficient and necessary for
rescalability of infinite models.We finally characterise the acceptable ways to handle uncertainty in infinite models
based on Cox's assumptions. Certain closure properties must be assumed before all the axioms of ordered fields are satisfied.
Once this is done, a proper plausibility model can be embedded in an ordered field containing the reals, namely either standard
probability (field of reals) for a real valued plausibility model, or extended probability (field of reals and infinitesimals)
for an ordered plausibility model.
Kanerva, P., Sjödin, G., Kristoferson, J., Karlsson, R., Levin, B., Holst, A., Karlgren, J., and Sahlgren, M.
Computing with large random patterns.
In Uesaka, Y., Kanerva, P,, and Asoh. H. (eds.),
Foundations of Real-World Intelligence (pp. 251-311). Stanford, Calif.: CSLI Publications, 2001.
Abstract: We describe a style of computing that differs from traditional numeric and symbolic computing and is suited for
modeling neural networks. We focus on one aspect of ``neurocomputing,'' namely, computing with large random patterns,
or high-dimensional random vectors, and ask what kind of computing they perform and whether they can help us understand
how the brain processes information and how the mind works. Rapidly developing hardware technology will soon be able to produce
the massive circuits that this style of computing requires. This chapter develops a theory on which the computing could be based.
Chapter V includes these articles:
Kanerva, P.
Analogy as a basis of computation.
(pp. 254-272)
Sjödin, G.
The Sparchunk Code: A method to build higher-level structures in a sparsely encoded SDM
(pp. 272-282)
Uesaka, Y., Kanerva, P,, and Asoh. H., eds.
Foundations of Real-World Intelligence.
Stanford, Calif.: CSLI Publications, 2001. (Distributed by the Chicago University Press.)
Book abstract: In 1992 Japan's Ministry of International Trade and Industry (MITI) began a research program in Real World Computing,
as a successor to the Fifth Generation Computing program of the previous decade, complementing the fifth-generation approach.
Its objective is to lay a foundation and to pursue the technical realisation of humanlike flexible and intelligent information processing.
This book collects results of ten years of original research by five research laboratories in Japan and Europe, whose research focus has
been the theoretical and algorithmic foundations of intelligence as manifested in the real world an in our dealing with it.
Real-world intelligent systems handle complex, uncertain, dynamic, multimodal information in real time. Both explicit and implicit
information are important. Hence we need to develop a novel integrated framework of representing knowledge and making inferences based in it.
It is impossible to pre-program all the knowledge needed for coping with the variety and complexity of real environments, and therefore learning
and adaptation are keys to intelligence. Learning is a kind of meta-programming strategy. Instead of writing programs for specific tasks, we must
write programs that modify themselves based on a system's interaction with its environment.
The book includes chapters on inference and learning with graphical models, approximate reasoning, evolutionary computation and beyond,
methodology of distributed and active learning, and computing with large random patterns. The treatment is mathematically rigorous, and
the discussion of issues is of general interest to an educated reader at large. The book provides excellent reading for graduate courses
in Computer Science, Cognitive Science, Artificial Intelligence, and Applied Statistics.
Yoshinori Uesaka is a professor of information sciences at the Science University of Tokyo. Pentti Kanerva is a senior researcher at the
Swedish Institute of Computer Science. Hideki Asoh is a senior researcher at the Electrotechnical Laboratory in Tsukuba City, Japan.
2002
Jussi Karlgren, Björn Gambäck, and Pentti Kanerva, (editors)
Conference arranged by SICS.
Acquiring (and Using) Linguistic (and World) Knowledge for Information Access or Theory for systems; application for theories.
Notes from AAAI Spring Symposium on Acquiring (and Using) Linguistic (and World) Knowledge for Information Access, Stanford University,
California. AAAI.
Abstract: To move forward the research frontier in the general field of information access, one of the bottlenecks we need to address
is understanding textual content somewhat better. While full text understanding remains a distant and possibly unattainable goal, advances
in content analysis beyond the simple word-occurrence statistics or name-recognition algorithms used today would seem to be desirable.
Current information-retrieval systems deliver results using a simple text and content model. Much better models are necessary for information
access tasks that involve information refinement, meaning tasks that involve processing information in text. Models that must either be adaptive
or easily adapted by some form of low-cost intervention; and that must support incremental knowledge build-up. The first requirement involves
acquisition of information from unstructured data; the second involves finding an inspectable and transparent model and developing an understanding
of knowledge-intensive interaction.
Douglas Wikström
A Note on the Malleability of the El Gamal Cryptosystem
To appear in Proceedings of Indocrypt 2002, LNCS Springer Verlag
Abstract: The homomorphic property of the El Gamal cryptosystem is useful in the construction of efficient protocols. It is believed that only
a small class of transformations of cryptotexts are feasible to compute. In the program of showing that these are the only computable transformations
we rule out a large set of natural transformations.
Douglas Wikström
The Security of a Mix-Center Based on a Semantically Secure
Cryptosystem
To appear in Proceedings of Indocrypt 2002, LNCS Springer Verlag
Abstract: We introduce a definition of a re-encryption mix-center, and a definition of security for such a mix-center. Then we prove that any
semantically secure public key system, which allows re-encryption, can be used to construct a secure mix-center.
|
|
Page Formatted for Printing