• Corpus ID: 5171233

Stochastic Learning Algorithms for Modeling Human Category Learning

@article{Matsuka2007StochasticLA,
  title={Stochastic Learning Algorithms for Modeling Human Category Learning},
  author={Toshihiko Matsuka and James E. Corter},
  journal={World Academy of Science, Engineering and Technology, International Journal of Computer, Electrical, Automation, Control and Information Engineering},
  year={2007},
  volume={1},
  pages={1170-1178}
}
  • T. Matsuka, J. Corter
  • Published 20 April 2007
  • Computer Science
  • World Academy of Science, Engineering and Technology, International Journal of Computer, Electrical, Automation, Control and Information Engineering
Most neural network (NN) models of human category learning use a gradient-based learning method, which assumes that locally-optimal changes are made to model parameters on each learning trial. This method tends to underpredict variability in individual-level cognitive processes. In addition many recent models of human category learning have been criticized for not being able to replicate rapid changes in categorization accuracy and attention processes observed in empirical studies. In this… 

Figures and Tables from this paper

Simple, individually unique, and context-dependent learning methods for models of human category learning
  • T. Matsuka
  • Computer Science
    Behavior research methods
  • 2005
TLDR
An alternative, hypothesis-testing, learning algorithm on the basis of a stochastic optimization method is introduced, termed SCODEL, which is the first modeling attempt to depict individually unique and context-dependent learning processes.
Generalized Exploratory Model of Human Category Learning Toshihiko
One problem in evaluating recent computational models of human category learning is that there is no standardized method for systematically comparing the models’ assumptions or hypotheses. In the
Generalized Exploratory Model of Human Category Learning
TLDR
In the present study, a flexible general model (called GECLE) is introduced that can be used as a framework to systematically manipulate and compare the effects and descriptive validities of a limited number of assumptions at a time.
On the Learning Algorithms of Descriptive Models of High-Order Human Cognition
TLDR
The present research attempts to bridge the gap by identifying five critical issues overlooked by previous modeling research and then introducing a modeling framework that addresses the issues and offers better qualitative plausibleness.

References

SHOWING 1-10 OF 20 REFERENCES
Generalized Exploratory Model of Human Category Learning
TLDR
In the present study, a flexible general model (called GECLE) is introduced that can be used as a framework to systematically manipulate and compare the effects and descriptive validities of a limited number of assumptions at a time.
SUSTAIN: A Model of Human Category Learning
TLDR
SUSTAIN's adaptive architecture allows it to master simple classification problems quickly, while still retaining the capacity to learn difficult mappings, as compared to other classification models.
Rule-plus-exception model of classification learning.
TLDR
The authors propose a rule-plus-exception model (RULEX) of classification learning, which accounts for numerous fundamental classification phenomena, including prototype and specific exemplar effects, sensitivity to correlational information, difficulty of learning linearly separables versus nonlinearly separable categories, selective attention effects, and difficulty oflearning concepts with rules of differing complexity.
A model of probabilistic category learning.
A new connectionist model (named RASHNL) accounts for many "irrational" phenomena found in nonmetric multiple-cue probability learning, wherein people learn to utilize a number of discrete-valued
Comparing modes of rule-based classification learning: A replication and extension of Shepard, Hovland, and Jenkins (1961)
TLDR
The main results mirrored those of Shepard et al., with the ordering of task difficulty being the same as in the original study, but a much richer data set was collected, which enabled the generation of block-by-block learning curves suitable for quantitative fitting.
Eyetracking and selective attention in category learning
Effect of relevance shifts in category acquisition: a test of neural networks.
  • S. Macho
  • Psychology
    Journal of experimental psychology. Learning, memory, and cognition
  • 1997
TLDR
Simulation of the results by the independentcue model, the configural cue model, and ALCOVE demonstrates that these models must be expanded by a mechanism that inhibits the integration of information acquired before and after the relevance shift.
Attention, similarity, and the identification-categorization relationship.
  • R. Nosofsky
  • Psychology
    Journal of experimental psychology. General
  • 1986
TLDR
A unified quantitative approach to modeling subjects' identification and categorization of multidimensional perceptual stimuli is proposed and tested and some support was gained for the hypothesis that subjects distribute attention among component dimensions so as to optimize categorization performance.
Correlation, conflict, and choice.
We examined the degree to which individuals adapt their decision processes to the degree of interattribute correlation and conflict characterizing a decision problem. On the basis of an
...
...