– A Deep Learning Tutorial by Richard Socher, Yoshua Bengio and Chris Manning.
– Selective Sharing for Multilingual Dependency Parsing. I always like the work from MIT people. This is an interesting paper on multilingual learning. The model does not assume the existing of parallel corpus, which makes it more practical. Moreover, it can transfer linguistic structures between unrelated languages.
– Unsupervised Morphology Rivals Supervised Morphology for Arabic MT. Very close to my thesis. In fact, I just need to read Mark Johnson’s 2009 paper Improving nonparameteric Bayesian inference: experiments on unsupervised word segmentation with adaptor grammars.