I am newbie in Machine learning. As we know, in machine learning the aim of an algorithm is to find out (approximate) the output from some examples as training example. these training examples help the algorithm to predict the output of other examples which have not occurred yet. According to Wikipedia, we need "inductive bias" by which we can be assured the algorithm can approximate the output correctly. But the thing is I am not sure that I catch the concept. does "inductive bias" mean we add another examples to our training set or we delimit the training examples? any precise explanation is appreciated.
Modified 7 years, 5 months ago
Viewed 173 times
$\begingroup$ Inductive bias might mean that you think that the training examples generalize. This is the same as scientific induction, in which you determine laws from experiments, and expect them to apply in other circumstances. $\endgroup$– Yuval FilmusOct 14, 2015 at 4:32
Add a comment |