Learning with Similarities on Subsets
Many machine learning models are based on similarities between new examples and previously observed examples. Such models are extremely flexible and can adapt to a wide range of tasks. However, if examples are composed of many variables, then even if we collect a large number of examples, it is possible that no two examples will be significantly similar. This, in turn, means that a learning algorithm may require an unreasonably large number of examples to...[Show more]
|Collections||Open Access Theses|
|Ruderman Thesis 2017.pdf||1.19 MB||Adobe PDF|
Items in Open Research are protected by copyright, with all rights reserved, unless otherwise indicated.