Similarity Measure Learning for Analogical Transfer – SMeLT

The aim of SMeLT is to provide a methodology to learn a
similarity measure that is optimized for a given analogical transfer task.
Among the different tasks that computational analogy systems
implement, the transfer task matches a predictive and hypothetical
inference in which some knowledge is extrapolated from a similar
situation in order to interpret a new situation and complete its
description.
By providing a set of quality indicators for the similarity measure,
and a metric learning method that optimizes these indicators,
this project will unlock a major bottleneck that currently prevents a widespread application of transfer
methods to real scenarii, which is to learn a similarity
measure that is adequate for the task at hand.
The project is a pluridisciplinary effort,
that brings together researchers from cognitive science,
computer science (computational analogy and similarity measures specialists),
and health sciences (medical decision support specialists).
The proposed methodology will be evaluated in two different application domains:
the cooking domain, and the domain of decision support for the therapeutic management of breast cancer.