Effective queries in machine learning -
i reading ml text has question:
if have supervisor can provide label x, should choose x learn fewer queries?
so interpret mean:
we asked expedite learning process selecting test elements provide discriminatory information.
i presume iterative algorithm: i.e. selection of succeeding test elements depends upon elements selected. not clear question is:
- would selection of succeeding elements also depend on results of prior queries.
so question here be: there information-theoretic unique answer above? or there multiple approaches may of similar effectiveness/speed?
** footnote: not homework - independent learning
there multiple approaches. in fact, setting machine learning task called active learning. there lot of methods choosing x based on different strategies - see section 3 of this survey paper.
Comments
Post a Comment