Computer Science – Learning
Scientific paper
Jan 2002
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2002iaf..confe.375e&link_type=abstract
IAF abstracts, 34th COSPAR Scientific Assembly, The Second World Space Congress, held 10-19 October, 2002 in Houston, TX, USA.,
Computer Science
Learning
Scientific paper
The ultimate aim of any system designed to analyse a signal for structured and intelligent-like behaviour, is not merely to provide a Boolean filter but to have the ability to deduce what category the signal fits - language, image, music noise etc - and then to decipher its contents. Up until now, my research has concentrated on the lower level universals of language structure and how they can be detected1,2. However, in this paper, I begin to look into developing techniques, which use only surface structure information deduced from the signal sample itself, for learning what underlies the higher level elements of the language (lingua ex machina) hierarchy, where syntax meets semantics. The aim therefore is to move towards developing techniques, which will unlock a signal's meaning using its internal cohesion and constraints. In doing so, I look at how the most disparate of human language orthographic forms mirror each other's underlying structure, despite their encoding strategies and how classes of words give their secrets away by the functional `friends' they keep. To address such linguistic problems, which reach down to and beyond the very core fundamental concepts of what we currently understand as language, I have had to develop new techniques to provide ways of learning and understanding syntactic behaviour from surface structure alone. Although each word or symbol is arbitrarily paired with its meaning, empirical evidence has shown that the semantic class or part-of-speech to which it belongs is constrained by its behaviour and according to underlying interactive rules. By using functional elements/words detected from lower level analysis and visualisation techniques for bonding, I have begun to develop algorithms, which group a communication's word classes, using the notion of polyposici and uniposicii behaviour within minimal constrained pairings. Initial trials with English confirm such behaviour and show how these semantic classes can be detected and words attributed their membership, without prior knowledge, using unsupervised positional rule-based criteria and statistical variance. Finally, I present techniques developed for non-language structured signals: - How the statistical `fingerprint' of a binary image, such as the recent challenge set by Dr Dutil3 and the Arecibo - Further analysis of the structure of music and how it differs from language. References 1. Elliott J, Atwell E &Whyte, W. 2001. First Stage Identification of Syntactic Elements in an Extraterrestrial 2. Elliott, John &Atwell, Eric, 1999. Language in signals: the detection of generic species-independent intelligent 3. Dutil, Y, 2001. University of Catalunya, Barcelone, Spain &ABB Bomem Inc, 4. Arecibo Observatory. http://www.naic.edu/.
No associations
LandOfFree
The Filtration of Inter-Galactic Objets Trouvés and the Identification of the Lingua Ex Machina Hierarchy does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with The Filtration of Inter-Galactic Objets Trouvés and the Identification of the Lingua Ex Machina Hierarchy, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and The Filtration of Inter-Galactic Objets Trouvés and the Identification of the Lingua Ex Machina Hierarchy will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-1331582