While a lot of progress has been made on efficient training with several loss functions, the problem of endowing learners with a mechanism for feature selection is still unsolved. Linear models have enjoyed great success in structured prediction in NLP. In the second half of the talk, I will talk about structured sparse modeling in structured prediction.
We show how first-order logical constraints can be handled efficiently, even though the corresponding subproblems are no longer combinatorial, and report experiments in dependency parsing, with state-of-the-art results. We sidestep that difficulty by adopting an augmented Lagrangian method (ADMM) that accelerates model consensus by regularizing towards the averaged votes. However, in cases where lightweight decompositions are not readily available (e.g., due to the presence of rich features or logical constraints), the original subgradient algorithm is inefficient. Dual decomposition has been recently proposed as a way of combining complementary models, with a boost in predictive power. In the first half of the talk, I will describe a new dual decomposition method for structured classification which is suitable for logically constrained problems, with applications in NLP.
0 kommentar(er)
