Robust Inference with Simple Cognitive Models

Developing theories of how information is processed to yield inductive inferences is a key step in understanding intelligence in humans and machines. Humans, across tasks as diverse as vision and decision making, appear to be extremely adaptive and successful in dealing with uncertainty in the world. Yet even a cursory examination of the books and journals covering machine learning reveals that this branch of AI rarely draws on the cognitive system as a source of insight. In this article I show how fast and frugal heuristics – cognitive process models of inductive inference – frequently outperform a wide selection of standard machine learning algorithms. This finding suggests a cognitive-inspired route toward robust inference in the context of meta-learning.

[1]  I. J. Myung,et al.  Toward a method of selecting among computational models of cognition. , 2002, Psychological review.

[2]  C. S. Wallace,et al.  Coding Decision Trees , 1993, Machine Learning.

[3]  Robert A. Wilson,et al.  Book Reviews: The MIT Encyclopedia of the Cognitive Sciences , 2000, CL.

[4]  Cleotilde Gonzalez,et al.  Instance-based learning in dynamic decision making , 2003 .

[5]  R. Nosofsky.  Relations between exemplar-similarity and likelihood models of classification , 1990 .

[6]  G Gigerenzer,et al.  Reasoning the fast and frugal way: models of bounded rationality. , 1996, Psychological review.

[7]  R. Schapire.  The Strength of Weak Learnability , 1990, Machine Learning.

[8]  P. Todd,et al.  Simple Heuristics That Make Us Smart , 1999 .

[9]  L. Macchi,et al.  Thinking : psychological perspectives on reasoning, judgment and decision making , 2005 .

[10]  C A Nelson,et al.  Learning to Learn , 2017, Encyclopedia of Machine Learning and Data Mining.

[11]  John R. Anderson.  The Adaptive Character of Thought , 1990 .

[12]  C. Bishop.  The MIT Encyclopedia of the Cognitive Sciences , 1999 .

[13]  H. Simon,et al.  Rational choice and the structure of the environment. , 1956, Psychological review.

[14]  P. Todd,et al.  More Is Not Always Better: The Benefits of Cognitive Limits , 2003 .

[15]  J. Rieskamp,et al.  SSL: a theory of how people learn to select strategies. , 2006, Journal of experimental psychology. General.

[16]  San Cristóbal Mateo,et al.  The Lack of A Priori Distinctions Between Learning Algorithms , 1996 .

[17]  Pat Langley,et al.  Elements of Machine Learning , 1995 .

[18]  A. Bröder,et al.  Take the best versus simultaneous feature matching: probabilistic inferences from memory and effects of representation format. , 2003, Journal of experimental psychology. General.

[19]  Dr. Marcus Hutter,et al.  Universal artificial intelligence , 2004 .

[20]  Raj Reddy,et al.  Foundations and grand challenges of artificial intelligence , 1988 .

[21]  I. Erev,et al.  On adaptation, maximization, and reinforcement learning among cognitive strategies. , 2005, Psychological review.

[22]  Marie desJardins,et al.  Evaluation and selection of biases in machine learning , 1995, Machine Learning.

[23]  R. Selten,et al.  Bounded rationality: The adaptive toolbox , 2000 .

[24]  Ricardo Vilalta,et al.  A Perspective View and Survey of Meta-Learning , 2002, Artificial Intelligence Review.

[25]  Eric J. Johnson,et al.  The adaptive decision maker , 1993 .

[26]  Jorma Rissanen,et al.  Fisher information and stochastic complexity , 1996, IEEE Trans. Inf. Theory.