The disembodied predictor stance

Argue that predictors are embodied, and predictions affect their own realization.Demonstrate a technical framework explicitly acknowledging prediction embodiment.Present application domains that can benefit from an embodied prediction viewpoint.Argue for the unity of science through the universal use of embodied predictors. Pattern recognition is typically described as the discipline investigating how to recognize patterns and regularities in data, with the description leaving tacit that these patterns and regularities are somehow exploited, applied, acted upon, or simply announced once recognized. The aforementioned omission is more than a linguistic one, and is reflected on the emphasis that technical, theoretical, and empirical work on pattern recognition places on the predictors it develops, analyzes, and deploys. Most research on pattern recognition adopts, effectively, a stance amounting to treating the predictors as being disembodied, taken to mean that they operate without affecting the environment about which they make predictions. This essay argues for the dismissal of this stance, and demonstrates that the adoption of an embodied predictor stance is philosophically and technically not only possible, but also desirable.

[1]  A. Clark.  Whatever next? Predictive brains, situated agents, and the future of cognitive science. , 2013, The Behavioral and brain sciences.

[2]  David Haussler,et al.  Decision Theoretic Generalizations of the PAC Model for Neural Net and Other Learning Applications , 1992, Inf. Comput..

[3]  Kousha Etessami,et al.  On the Complexity of Nash Equilibria and Other Fixed Points , 2010, SIAM J. Comput..

[4]  N. Littlestone.  Learning Quickly When Irrelevant Attributes Abound: A New Linear-Threshold Algorithm , 1987, 28th Annual Symposium on Foundations of Computer Science (sfcs 1987).

[5]  Linda Sellie,et al.  Toward efficient agnostic learning , 1992, COLT '92.

[6]  H. Simon,et al.  American Association for Public Opinion Research Bandwagon and Underdog Effects and the Possibility of Election Predictions , 2009 .

[7]  Robert Nozick,et al.  Newcomb’s Problem and Two Principles of Choice , 1969 .

[8]  Loizos Michael,et al.  Introspective Forecasting , 2015, IJCAI.

[9]  Tom Fawcett,et al.  "In vivo" spam filtering: a challenge problem for KDD , 2003, SKDD.

[10]  Wei Chu,et al.  A contextual-bandit approach to personalized news article recommendation , 2010, WWW '10.

[11]  H. Robbins.  Some aspects of the sequential design of experiments , 1952 .

[12]  E. Sperner.  Neuer beweis für die invarianz der dimensionszahl und des gebietes , 1928 .

[13]  J. Fodor.  Special sciences (or: The disunity of science as a working hypothesis) , 1974, Synthese.

[14]  Loizos Michael.  Partial observability and learnability , 2010, Artif. Intell..

[15]  Simon M. Lucas,et al.  A Survey of Monte Carlo Tree Search Methods , 2012, IEEE Transactions on Computational Intelligence and AI in Games.

[16]  R. Merton.  The unanticipated consequences of purposive social action , 1936 .

[17]  T Brewin,et al.  Primum non nocere? , 1994, Lancet.

[18]  Lawrence A. Shapiro.  Embodied Cognition , 2010 .

[19]  D. Angluin.  Queries and Concept Learning , 1988 .

[20]  Leslie G. Valiant,et al.  A theory of the learnable , 1984, STOC '84.

[21]  Leslie G. Valiant,et al.  Robust logics , 1999, Symposium on the Theory of Computing.

[22]  Teresa Scantamburlo,et al.  How Mature Is the Field of Machine Learning? , 2013, AI*IA.

[23]  Loizos Michael,et al.  Simultaneous Learning and Prediction , 2014, KR.

[24]  Loizos Michael,et al.  Causal Learnability , 2011, IJCAI.

[25]  L. Brouwer.  Über Abbildung von Mannigfaltigkeiten , 1911 .

[26]  Marcello Pelillo,et al.  Similarity-Based Pattern Analysis and Recognition , 2013, Advances in Computer Vision and Pattern Recognition.

[27]  Michael R. Genesereth,et al.  General Game Playing: Overview of the AAAI Competition , 2005, AI Mag..

[28]  R. Merton.  The Self-Fulfilling Prophecy , 1948 .