Statistical measures for defining curriculum scoring function

Show simple item record Sadasivan, Vinu Sankar Dasgupta, Anirban 2021-03-16T12:19:00Z 2021-03-16T12:19:00Z 2021-02
dc.identifier.citation Sadasivan, Vinu Sankar and Dasgupta, Anirban, "Statistical measures for defining curriculum scoring function", arXiv, Cornell University Library, DOI: arXiv:2103.00147, Feb. 2021. en_US
dc.description.abstract Curriculum learning is a training strategy that sorts the training examples by some measure of their difficulty and gradually exposes them to the learner to improve the network performance. In this work, we propose two novel curriculum learning algorithms, and empirically show their improvements in performance with convolutional and fully-connected neural networks on multiple real image datasets. Motivated by our insights from implicit curriculum ordering, we introduce a simple curriculum learning strategy that uses statistical measures such as standard deviation and entropy values to score the difficulty of data points for real image classification tasks. We also propose and study the performance of a dynamic curriculum learning algorithm. Our dynamic curriculum algorithm tries to reduce the distance between the network weight and an optimal weight at any training step by greedily sampling examples with gradients that are directed towards the optimal weight. Further, we also use our algorithms to discuss why curriculum learning is helpful.
dc.description.statementofresponsibility by Vinu Sankar Sadasivan and Anirban Dasgupta
dc.language.iso en_US en_US
dc.publisher Cornell University Library en_US
dc.title Statistical measures for defining curriculum scoring function en_US
dc.type Pre-Print en_US
dc.relation.journal arXiv

Files in this item

Files Size Format View

There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record

Search Digital Repository


My Account