AAAI 2008 Fall Symposium Series
Arlington, Virginia — November 5–7, 2009
The challenge of designing a human-level learner is central to creating a real-life computational equivalent of the human mind. It demands the level of robustness and flexibility of learning that today is available in biological systems only. Therefore, it is essential that we better understand at a computational level how biological systems naturally develop their cognitive and learning functions. In recent years, biologically inspired cognitive architectures (BICA) have emerged as a powerful new approach toward gaining this kind of understanding. The impressive success of BICA-2008 was clear evidence of this trend. As the second event in the series, BICA-2009 continues our attack on the challenge, with the overall atmosphere of excitement and potential, brainstorming and collaboration.
Topics
– Bridging the gap between AI and biology: robustness, flexibility, integrity.
– BICA models of learning: bootstrapped, self-regulated (SRL), meta-learning.
– Scalability, limitations and ‘critical mass’ of cognitive vs. subcognitive learning.
– Biological constraints vital for learning.
– Physical support of conscious experience.
– Formal theory of cognitive architectures.
– Emotional feelings and values in artifacts.
– Measuring minds of machines and humans.
Symposium Focus and Spirit
The challenge addressed by this symposium is stated above. The narrow focus is on the idea to replicate in artifacts the phenomenon of natural cognitive growth (human-like learning and cognitive development), using models of learning borrowed from biology, neuroscience, cognitive / developmental psychology, cognitive linguistics, educational and social sciences. Specific tasks include: to identify critical components of human-like learning mechanisms that enable transformative cognitive growth in BICA; to understand at a computational level the leverage of biological constraints in self-regulated cognitive growth; to design curricula, tests and scalability metrics for artifacts and a roadmap to solving the challenge.
The spirit of the symposium is science (and, indeed, its focus is on a fundamental scientific problem). This symposium is not about the closed DARPA program or its successors, it is not a formal presentation event, not a publishing venue, not a funding opportunity forum, and not an industry teaming day (while all these elements may be present in it to some extent). It is a working seminar where researchers come together with new ideas and have a discussion.
Therefore, the majority of presentations will be short and exciting, while longer talks will be used to set the stage for discussion panels (see the Format section below).
The three tasks listed above (the ‘critical mass’, vital constraints and scalability metrics) will underlie the objectives of one or two discussion panels. In addition, a discussion panel on Emotions in Artifacts is organized jointly by Drs. Eva Hudlicka (Principal Scientist and President, Psychometrix Associates) and Rodrigo Ventura (Institute for Systems and Robotics, Lisbon, Portugal). Another discussion panel will follow the Joint Session on Future Funding of Research in Learning Technologies and will continue its topic (please see http://binf.gmu.edu/~asamsono/bica/bica2009mces2009.pdf). Additional discussion panels are currently under consideration, and proposals are welcome (please see http://binf.gmu.edu/~asamsono/bica/bica2009panels.pdf).
More generally and apart from the discussion panels, the symposium is focused on developing an understanding of how the machinery of the brain-mind develops the abilities to control attention, perceive objects and events, understand situations and mental states, reason about the past, plan for the future, learn from experience, and as a result – grow cognitively from a baby to an adult, becoming capable of the execution of purposeful behavior in the world. Therefore, the symposium will bring together recent studies in artificial intelligence, in psychological and brain sciences, including cognitive modeling, cognitive robotics and intelligent agent studies, models and theories of self-regulated learning, language acquisition and bootstrapped learning – all in the context of the selected challenge.
In terms of categories of research, the first priority of this symposium is to showcase recent modeling and rapid prototyping experience aimed at building architectures of cognitive agents that have been inspired by the human brain and, in a definite sense, operate like the human mind. At the same time, theoretical discussion of the underlying mechanisms is equally encouraged. In addition, the symposium will address potential targets of applications of the BICA technology, including cognitive robotics, machine learning, natural language processing, intelligent social and pedagogical agents, and human-computer interface.
Broader Scope and Topics
Submissions are not expected to be limited to the narrow focus outlined above, and should cover domains relevant to the challenge. Therefore, tentative topics of the technical sessions of the symposium include (yet are not limited to) the following list (please also see the list of key speakers on page 7). Bridging the gap between natural and artificial intelligence: robustness, flexibility, integrity Models of natural cognitive growth: self-regulation, bootstrapping, meta-learning Critical components of human-like learning that enable transformative cognitive growth Vital biological constraints informed by neuroscience and their leverage in learning systems Cognitive vs. subcognitive forms of learning: scalability laws and metrics for growing BICA.
Physical support of conscious experience: the emergent Self and self-awareness in artifacts Formal theory of cognitive and metacognitive architectures and their natural development Language acquisition, symbol grounding, and the ‘critical mass’ of a universal learner Reading and measuring minds of machines and humans: the second cognitive revolution The origin and the function of emotional feelings and values in humans and in artifacts.
More information at: http://members.cox.net/bica2009/
