Ronan Collobert Jason Weston Leon Bottou Michael Karlen Koray Kavukcouglu Pavel Kuksa Common Approaches in NLP Using Task-specific features Knowledge injection about structure of data Expertise from Linguists Approach used in the paper No task-specific feature engineering Minimal prior Knowledge Part-Of-Speech (POS) Tagging Syntactic Parsing Chunking Shallow Parsing Named Entity Recognition Person, Location etc. Semantic Role Labelling Words to Feature Vectors Look up Table Random initialization vs Unsupervised Pre- training Extending to any Discrete Features Extracting Higher level Features from Word Feature Vectors Window Approach Sentence Approach Word-Level Log Likelihood Only words are taken independently for optimizing the weights Sentence-Level Log Likelihood Optimization function takes into account all the tags as well as transitions between tags Stochastic Gradient Standard Optimization Algorithm