**1. Midterm Presentation** (Please select your topic from the following list)

  * Convex Optimization: {{:convex_opt_tom_paper.pdf|paper1}}

  * Graphical Models: {{:graphical_models.pdf|paper2}}

  * Deep Neural Networks: {{:learning-deep-ai.pdf|paper3}} **(by Hong)**

  * Feature Dimensionality Reduction: {{:dimred-chris.pdf|paper4}} **(by Boze)**

  * Conditional Random Field: {{:crf.pdf|paper5}} and {{:crf-jafferty2001.pdf|paper6}} **(by Martin)**

  * Stochastic Gradient Descent: {{:bottou-mlss-2003.pdf|paper7}}, {{:bottou-onlinelearning-98.pdf|paper8}} and {{:nips2011_distributed_optimization.pdf|paper9}}  **(by Hengyue)**


Each person makes a presentation (35-40 minutes) on one of the above topics and leaves 5 minutes for Q&A. **The midterm presentation has been scheduled in a single 3-hr time slot from 11am-2pm at __LAS3033__ (not our regular classroom) on Oct 29 (Mon). As a result, the class on Oct 24 is cancelled.** 



**2. Final Presentation:**  (Please select your topic from the following list or let me know any other topic you may be interested in.)

  * Discriminative Training of HMMs:  {{:jiang_dt_survey.pdf|paper10}} and {{:woodland_mmie.pdf|paper11}} **(by Hengyue)**

  * Bayesian Learning of HMMs:  {{:gauvain_map.pdf|paper12}} and {{:huo_online.pdf|paper13}}

  * Latent Semantic Analysis for Language Model:  {{:belleg_lsa.pdf|paper14}} **(by Boze)**

  * Weighted Finite State Transducer for Speech and Language Processing:  {{:wfst-lvcsr.pdf|paper15}} **(by Hong)**

  * SPAM filtering:  {{:spam-filtering.pdf|paper16}}

  * Auto Summarization: {{:auto-summarization.pdf|paper17}}  **(by Martin)**

  * Open-Domain Q&A: {{:ibm-ai-qna.pdf|paper18}}

Similar to the midterm presentations, each person makes a presentation (35-40 minutes) on one of the above topics and leaves 5 minutes for Q&A. The final presentation has been scheduled in a single 3-hr time slot from **2-5pm on Dec 14 (Fri), location is LAS3033**. As a result, two classes on Nov 28 and Dec 3 are cancelled.