Introduction
Welcome to the home page of Ramesh Maruthi Nallapati. I am a post doctoral fellow at the Machine Learning department in the School of Computer Science of Carnegie Mellon University, working with Prof. John Lafferty and Prof. William Cohen. I also work very closely with Prof. Eric Xing. My research interests are in Machine Learning, and its applications to Information Retrieval, Information Extraction, and Data mining. Here is my detailed CV
Latest News
I just accepted a position as a Research Associate at the CS department of Stanford University with Prof. Chris Manning. I will be starting my new job on April 7th. More news to follow.
As a post-doctoral fellow at CMU, I have been working in the framework of an unsupervised technique for document analysis called latent topic modeling.
Recently, we built a model called topic tomography, that can automatically analyze evolution of topic content with time at various time-scales of resolution, and published this work in KDD 2007. A couple of demos based on my KDD 2007 paper are here and here. A video lecture of my talk at KDD on this work can be found here.
I also recently implemented parallel and distributed versions of the variational inference algorithm for the Latent Dirichlet Allocation Model. This work is now available as a paper at ICDM workshop on high performance data mining.
In Fall 2007), I assisted Prof. Eric Xing in the Probabilistic Graphical Models course on a voluntary basis and I have been enjoying it very much. Here is a set of slides I made on the Junction Tree Algorithm (ppt format). Here are some slides I made on Conditional Random Fields (ppt format).
I also recently gave a guest lecture in Prof. William Cohen’s course on Social Media Analysis on the topic of topic models for community analysis.
Background
I earned a Ph.D. from the Computer Science department at University of Massachusetts, Amherst in 2006, where I worked with Prof. James Allan at the Center for Intelligent Information Retrieval. Previously, I earned a Master’s in Mechanical Engineering at the same school where I worked with Prof. Blair Perot. I earned my undergrad degree in Mechanical Engineering from Indian Institute of Technology, Mumbai, India, in 1998. A more detailed CV is here.
Contacts
I may be reached on email at [nmramesh]at[cs]dot[cmu]dot[edu]. My office is 4301E, Doherty Hall and my phone number is 412-268-3936.
Publications
2008
· Joint Topic Models for Text and Citations, Nallapati, R., Ahmed, A., Xing, E.P., Cohen, W., submitted to KDD 2008 [pdf]
· Exploiting Feature hierarchy for Transfer Learning in named Entity Recognition, Arnold, A., Nallapati, R. and Cohen W., ACL, 2008 [pdf]
· Link-PLSA-LDA: A new unsupervised model for topics and influence in blogs, Nallapati, R. and Cohen W., International Conference for Weblogs and Social Media, 2008, [pdf]
2007
·
Multi-scale
Topic Tomography, Nallapati, R., Cohen, W., Ditmore, S., Lafferty, J., Ung,
K., ACM-KDD, 2007. [pdf] [ppt]
· Parallelized Variational EM for Latent Dirichlet Allocation: An experimental evaluation of speed and scalability, Nallapati, R., Cohen, W., Lafferty, J., ICDM workshop on high performance data mining, 2007 [pdf] [ppt]
· Sparse Word Graphs: A scalable algorithm for capturing word correlations in topic models, Nallapati, R., Ahmed, A., Cohen W., Xing, E., ICDM workshop on High performance data mining, 2007 [pdf] [ppt]
· A Comparative Study of Methods for Transductive Transfer Learning, Arnold, A., Nallapati, R., and Cohen, W., ICDM 2007 Workshop on Mining and Management of Biological Data, 2007 [pdf]
2006
·
Smoothed
Dirichlet distribution: Understanding Cross-entropy ranking in information
retrieval, Ph.D. thesis,
·
The
Smoothed Dirichlet distribution: A new Building block for generative topical
models, Nallapati, R., Minka, T. and Robertson S., CIIR Tech Report [pdf]
·
The
Smoothed Dirichlet distribution: Explaining KL-divergence in information
retrieval, Nallapati R., Minka, T.,
2004
·
Event Threading
within News Topics, Nallapati, R., Feng, A., Peng, F., Allan, J., ACM-CIKM,
2004. [ps] [pdf]
· Discriminative Models for Information Retrieval, Nallapati, R., ACM-SIGIR, 2004. [ps] [pdf]
2003
· Relevant Query Feedback in Statistical Language Modeling, Nallapati, R., Croft, W.B. and Allan, J., short-paper, ACM-CIKM, 2003. [ps] [pdf]
· An Adaptive Local Dependency language Model: Relaxing the Naïve Bayes’ Assumption , Nallapati, Ramesh and Allan, J. , Workshop on Mathematical/Formal Methods in Information Retrieval, SIGIR-2003. [ps] [pdf]
· Semantic Language Models for Topic Detection and Tracking, Nallapati, R, HLT-NAACL student research workshop, 2003. [ps] [pdf]
2002
· Capturing Term Dependencies using a Sentence Tree based Language Model, Nallapati, R. and Allan, J., ACM-CIKM ’02 conference, pp. 383-390. [ps] [pdf]
· UMass at TDT 2002, James Allan, Victor Lavrenko and Ramesh Nallapati, Proceedings Topic Detection and Tracking workshop, 2002. [pdf]
OTHER TECHNICAL REPORTS
· Evaluating the Quality of Query refinement Suggestions in Information Retrieval , Nallapati, R. and Shah, C., CIIR Technical Report IR-521, 2006 [pdf].
· Extraction of key-words from news stories, Ramesh Nallapati, James Allan and Sridhar Mahadevan, CIIR Technical report # IR-345. [pdf]
Mechanical Engineering (My old avatar)
· Conservation properties of unstructured staggered mesh methods, Perot B. and Nallapati, R., American Physical Society, 52nd Annual Meeting of the Division of Fluid Dynamics, Vol 44, No. 8, Nov 1999.
· Numerical simulation of free surface flows using a moving mesh, Perot B. and Nallapati, R., American Society of Mechanical Engineers: 2000 Fluids Engineering Summer Conference.