Output details
11 - Computer Science and Informatics
Queen Mary University of London
Differential Use of Implicit Negative Evidence in Generative and Discriminative Language Learning
<22>This is the first work to apply the well-known generative vs. discriminative distinction in machine learning to human learning. We show that a prominent debate in language acquisition (how children learn exceptions to grammatical rules) can be resolved through the different formal assumptions underlying discriminative vs. generative learning. We were the only cognitive science work to be invited to give an oral presentation at Neural Information Procession Systems conference, 2009, (22 oral presentations were awarded that year out of 1105 submissions). The work led to a grant from the Pattern Analysis, Statistical Modelling and Computational Learning Network 2 (PASCAL2, 2011, £45,000).