Neural-Symbolic Cognitive Reasoning - ReadingSample

The construction of robust computational models integrating reasoning and learning is a key research challenge for artificial intelligence. Recently, this challenge was also put forward as a fundamental problem in computer science [255]. Such a challenge intersects with another long-standing entry in the research agenda of artificial intelligence: the integration of its symbolic and connectionist paradigms. Such integration has long been a standing enterprise, with implications for and applications Further, the importance of efforts to bridge the gap between the connectionist and symbolic paradigms of artificial intelligence has also been widely recognised (see e. Valiant [255] has pointed out that the construction of rich computational cog-nitive models is one of the challenges computer science faces over the next few decades. A positive answer to this challenge would provide a characterisation of a semantics for cognitive computation, 1 as follows: The aim here is to identify a way of looking at and manipulating commonsense knowledge that is consistent with and can support what we consider to be the two most fundamental aspects of intelligent cognitive behaviour: the ability to learn from experience, and the ability to reason from what has been learned. We are therefore seeking a semantics of knowledge that can computationally support the basic phenomena of intelligent behaviour [255]. Valiant also described the characteristics of the semantic formalisation needed for supporting learning and reasoning: One set of requirements [for a semantics to be adequate for commonsense reasoning] is that it should support integrated algorithms for learning and reasoning that are computationally 1 The article [255] was published in 2003 in the Journal of the ACM, in a special issue celebrating its 50th anniversary. In that issue, the editor-in-chief at the time (J.Y. Halpern) invited winners of the Turing Award and Nevanlinna Prize to discuss up to three problems that these prominent researchers thought would be major challenges for computer science in the next 50 years.