Combining Evidence of Deep and Surface Similarity

Fisher, Douglas H.; Yoo, Jungsoon P.

doi:10.1016/B978-1-55860-200-7.50013-1

Combining Evidence of Deep and Surface Similarity

Douglas H. Fisher, Jungsoon P. Yoo

ICML 1991 pp. 46-50

doi:10.1016/B978-1-55860-200-7.50013-1 /icml/1991/fisher1991icml-combining/

Abstract

The traditional emphasis on surface similarity provides an incomplete account of how humans retrieve and categorize problem-solving experiences. Rather, ‘deep’ properties – those that are not immediately apparent but that can be inferred from the observables – play a critical role in memory activities, particularly those of experts. This paper provides a speculative (Hall & Kibler, 1985) account of the identification, inference, and use of deep features during categorization. In particular, we describe a machine learning system that categorizes problem-solving experiences and that retrieves past experiences that appear relevant to new situations; this retrieval is based on a combination of deep and surface similarties between the target and source problems.

ICML Semantic Scholar

Cite

Text

Fisher and Yoo. "Combining Evidence of Deep and Surface Similarity." International Conference on Machine Learning, 1991. doi:10.1016/B978-1-55860-200-7.50013-1

Markdown

[Fisher and Yoo. "Combining Evidence of Deep and Surface Similarity." International Conference on Machine Learning, 1991.](https://mlanthology.org/icml/1991/fisher1991icml-combining/) doi:10.1016/B978-1-55860-200-7.50013-1

BibTeX

@inproceedings{fisher1991icml-combining,
  title     = {{Combining Evidence of Deep and Surface Similarity}},
  author    = {Fisher, Douglas H. and Yoo, Jungsoon P.},
  booktitle = {International Conference on Machine Learning},
  year      = {1991},
  pages     = {46-50},
  doi       = {10.1016/B978-1-55860-200-7.50013-1},
  url       = {https://mlanthology.org/icml/1991/fisher1991icml-combining/}
}