Shimon Edelman

232 Uris Hall
Cornell University
Ithaca, NY
(607) 255-6365

The unique intellectual challenges posed by psychology and the neurosciences stem ultimately from the tremendous complexity of the prime mover of individuals and societies: the human brain. My current research focuses on one of the central puzzles of cognition: the manner in which brains deal with structured information. Structure-related tasks are ubiquitous in cognition: in language, for example, they arise from the compositional nature of the medium (speech) and the message (thought). Likewise, in vision, scenes are perceived as composed of recurring objects (and objects of parts), rather than as whole, indivisible entities.

The production and the assimilation of structured cues by the brain are constrained by the neuronal architecture, interconnection patterns and function, as well as by general principles of information processing, studied in engineering and computer science. In my work, therefore, I attempt to combine insights from these different disciplines, aiming, for each phenomenon at hand, at a model that would be couched in abstract computational (information-processing) terms, yet would be compatible with the basic neural-architectural constraints, and with behavioral data obtained from human subjects in carefully designed experiments. This approach is exemplified by an ongoing study of the processing of structure in vision and language, which has identified common statistical principles (such as co-occurrence of stimuli) that the brain uses in learning structured representations, and has cast them into biologically relevant computational models, which are being tested in behavioral experiments.

My next book, which paints the "big picture" about the mind/brain in computational terms, will be published by Oxford University Press in the summer of 2008.

representative recent publications

Zach Solan, David Horn, Eytan Ruppin and Shimon Edelman

in Proc. Natl. Acad. Sci. (102:11629-11634, August 16, 2005)

We address the problem, fundamental to linguistics, bioinformatics and certain other disciplines, of using corpora of raw symbolic sequential data to infer underlying rules that govern their production. Given a corpus of strings (such as text, transcribed speech, chromosome or protein sequence data, sheet music, etc.), our unsupervised algorithm recursively distills from it hierarchically structured patterns. The ADIOS (Automatic DIstillation of Structure) algorithm relies on a statistical method for pattern extraction and on structured generalization, two processes that have been implicated in language acquisition. It has been evaluated on artificial context-free grammars with thousands of rules, on natural languages as diverse as English and Chinese, and on protein data correlating sequence with function. This is the first time an unsupervised algorithm is shown capable of learning complex syntax, generating grammatical novel sentences, and proving useful in other fields that call for structure discovery from raw data, such as bioinformatics.

Unsupervised Learning of Visual Structure

Shimon Edelman, Nathan Intrator and Judah S. Jacobson

in Proc. 2002 International Workshop on Biologically Motivated Computer Vision, Tübingen, Germany

To learn a visual code in an unsupervised manner, one may attempt to capture those features of the stimulus set that would contribute significantly to a statistically efficient representation (as dictated, e.g., by the Minimum Description Length principle). Paradoxically, all the candidate features in this approach need to be known before statistics over them can be computed. This paradox may be circumvented by confining the repertoire of candidate features to actual scene fragments, which resemble the ``what+where'' receptive fields found in the ventral visual stream in primates. We describe a single-layer network that learns such fragments from unsegmented raw images of structured objects. The learning method combines fast imprinting in the feedforward stream with lateral interactions to achieve single-epoch unsupervised acquisition of spatially localized features that can support systematic treatment of structured objects.

Most of my publications are online at http://kybele.psych.cornell.edu/~edelman/archive.html, and so is my curriculum vitae.