Robot Baby 2001

In this paper we claim that meaningful representations can be learned by programs, although today they are almost always designed by skilled engineers. We discuss several kinds of meaning that representations might have, and focus on a functional notion of meaning as appropriate for programs to learn. Specifically, a representation is meaningful if it incorporates an indicator of external conditions and if the indicator relation informs action. We survey methods for inducing kinds of representations we call structural abstractions. Prototypes of sensory time series are one kind of structural abstraction, and though they are not denoting or compositional, they do support planning. Deictic representations of objects and prototype representations of words enable a program to learn the denotational meanings of words. Finally, we discuss two algorithms designed to find the macroscopic structure of episodes in a domain-independent way.

[1]  E. Rosch,et al.  Family resemblances: Studies in the internal structure of categories , 1975, Cognitive Psychology.

[2]  William P. Alston,et al.  Knowledge and the Flow of Information , 1985 .

[3]  James F. Allen.  An Interval-Based Representation of Temporal Knowledge , 1981, IJCAI.

[4]  David Chapman,et al.  Pengi: An Implementation of a Theory of Activity , 1987, AAAI.

[5]  Dana H. Ballard,et al.  Reference Frames for Animate Vision , 1989, IJCAI.

[6]  Ramanathan V. Guha,et al.  Building Large Knowledge-Based Systems: Representation and Inference in the Cyc Project , 1990 .

[7]  Fred I. Dretske.  Explaining Behavior: Reasons in a World of Causes , 1990 .

[8]  Ramanathan V. Guha,et al.  Cyc: toward programs with common sense , 1990, CACM.

[9]  Paul R. Cohen,et al.  Neo: learning conceptual knowledge by sensorimotor interaction with an environment , 1997, AGENTS '97.

[10]  Benjamin Kuipers,et al.  Map Learning with Uninterpreted Sensors and Effectors , 1995, Artif. Intell..

[11]  Rajesh P. N. Rao,et al.  Embodiment is the foundation, not a level , 1996, Behavioral and Brain Sciences.

[12]  Paul R. Cohen,et al.  Intelligent Support for Exploratory Data Analysis , 1998 .

[13]  Paul R. Cohen,et al.  Applicability of Reinforcement Learning , 1998 .

[14]  Paul R. Cohen,et al.  Interaction with a mixed-initiative system for exploratory data analysis , 1998, Knowl. Based Syst..

[15]  Paul R. Cohen,et al.  What are contentful mental states? Dretske's theory of mental content viewed in the light of robot learning and planning algortihms , 1999, AAAI/IAAI.

[16]  P. Bloom.  How children learn the meanings of words , 2000 .

[17]  Benjamin Kuipers,et al.  The Spatial Semantic Hierarchy , 2000, Artif. Intell..

[18]  Paul R. Cohen,et al.  A Method for Clustering the Experiences of a Mobile Robot that Accords with Human Judgments , 2000, AAAI/IAAI.

[19]  Paul R. Cohen,et al.  Identifying qualitatively different outcomes of actions: gaining autonomy through learning , 2000, AGENTS '00.

[20]  Paul R. Cohen,et al.  Multivariate Clustering by Dynamics , 2000, AAAI/IAAI.

[21]  Paul R. Cohen,et al.  Grounding knowledge in sensors: unsupervised learning for language and planning , 2001 .

[22]  Paul R. Cohen,et al.  Fluent Learning: Elucidating the Structure of Episodes , 2001, IDA.

[23]  Paul R. Cohen,et al.  An Algorithm for Segmenting Categorical Time Series into Meaningful Episodes , 2001, IDA.

[24]  Paul R. Cohen,et al.  Sequence Learning via Bayesian Clustering by Dynamics , 2001, Sequence Learning.