Joel Pfeiffer
Joseph J. Pfeiffer III
jpfeiffer at purdue dot edu

Research
I am a last year PhD student at Purdue University, studying machine learning in relational domains with Professor Jennifer Neville. I have interests in relational learning, particularly in partially observed network domains. I also study generative graph models for understanding how large networks form, and using them for anomaly detection. In addition, I am interested how incentivization can affect network sharing behavior, and how companies can use this to reach out to a large user base. Previously, I worked on robotics, where I worked on localization of objects when grasping.
Recent News
05/20: WWW 2015. (Our Paper)  "Overcoming relational learning biases to accurately predict preferences in large scale networks" 04/17: Paper accepted at SIGIR2015! 01/16: Paper accepted at WWW2015! 02/03: Talk at University of Arizona (Topic TBD) 01/28: AAAI 2015. (Our Paper)  "Incorporating Assortativity and Degree Dependence into Scalable Network Models." 01/13: Talk at Microsoft Research (Redmond)  "Learning and sampling from scalable generative graph models." 12/16: ICDM 2014. (Our Paper)  "Composite Likelihood Data Augmentation for WithinNetwork Statistical Relational Learning." 12/15: ICDM 2014. (Our Paper)  "A Scalable Method for Exact Sampling from Kronecker Models" Internships
Microsoft Research (Summer 2013)
Topic: Active Learning Mentors: Paul Bennett and Max Chickering Lawrence Livermore National Labs (Fall 2012)
Topic: Temporal graph modeling and attributed graph modeling Mentor: Brian Gallagher LivingSocial (Summer 2012)
Topic: Incentivized sharing in consumer websites Mentor: Elena Zheleva NASA JSC (20052011, 8 tours)
Topic: Robotic grasping, planning, etc. Mentor: Robert Platt, Jodi Graf, Tam Ngo, Robert Hirsh, Asher Lieberman Publications
Modeling Website Topic Cohesion at Scale to Improve Webpage Classification
Dhivya Eswaran, Paul N. Bennett and Joseph J. Pfeiffer IIIIn Proceedings of 38th Annual ACM SIGIR Conference (SIGIR 2015) [BibTeX] Overcoming relational learning biases to accurately predict preferences in large scale networks
Joseph J. Pfeiffer III, Jennifer Neville and Paul N. BennettIn Proceedings of 24th International World Wide Web Conference (WWW 2015) [PDF] [BibTeX] Incorporating Assortativity and Degree Dependence into Scalable Network Models
Stephen Mussman, John Moore, Joseph J. Pfeiffer III and Jennifer Neville In Proceedings of the 29th AAAI Conference on Artificial Intelligence (AAAI 2015) [PDF] [BibTeX] Composite Likelihood Data Augmentation for WithinNetwork Statistical Relational Learning
Joseph J. Pfeiffer III, Jennifer Neville and Paul Bennett In Proceedings of the 14th IEEE International Conference on Data Mining (ICDM 2014) [PDF] [Slides] [BibTeX] A Scalable Method for Exact Sampling from Kronecker Models
Sebastian Moreno, Joseph J. Pfeiffer III, Jennifer Neville and Sergey Kirshner In Proceedings of the 14th IEEE International Conference on Data Mining (ICDM 2014) [PDF] [BibTeX] Active Exploration in Networks: Using Probabilistic Relationships for Learning and Inference
Joseph J. Pfeiffer III, Jennifer Neville and Paul Bennett In Proceedings of the 23rd ACM International Conference on Information and Knowledge Management (CIKM 2014) [PDF] [BibTeX] Attributed Graph Models: Towards the Sharing of Relational Data
Joseph J. Pfeiffer III, Sebastian Moreno, Timothy La Fond, Jennifer Neville and Brian Gallagher KDD at Bloomberg, 2014 [PDF] [BibTeX] Assortativity in Chung Lu Random Graph Models
Stephen Mussman, John Moore, Joseph J. Pfeiffer III and Jennifer Neville In Proceedings of the 8th Workshop on Social Network Mining and Analysis (SNAKDD 2014) [PDF] [BibTeX] Attributed Graph Models: Modeling network structure with correlated attributes
Joseph J. Pfeiffer III, Sebastian Moreno, Timothy La Fond, Jennifer Neville and Brian Gallagher In Proceedings of the 23rd International World Wide Web Conference (WWW 2014), 2014 [PDF] [BibTeX] [Datasets] [Presentation  PDF] Combining Active Sampling with Parameter Estimation and Prediction in Single Networks
Joseph J. Pfeiffer III, Jennifer Neville and Paul N. Bennett In Proceedings of the ICML Structured Learning Workshop, 2013 [PDF] [BibTeX] Incentivized Sharing in Social Networks
Joseph J. Pfeiffer III and Elena Zheleva In Proceedings of the First International Workshop on Online Social Systems, 2012 [PDF] [BibTeX] Fast Generation of Large Scale Social Networks While Incorporating Transitive Closures
Joseph J. Pfeiffer III, Timothy La Fond, Sebastian Moreno and Jennifer Neville In Proceedings of the Fourth ASE/IEEE International Conference on Social Computing, 2012 [PDF] [BibTeX] Active Sampling of Networks
Joseph J. Pfeiffer III, Jennifer Neville and Paul N. Bennett In Proceedings of the 10th Workshop on Mining and Learning with Graphs, 2012 [PDF] [BibTeX] [PPT] Methods to Determine Node Centrality and Clustering in Graphs with Uncertain Structure
Joseph J. Pfeiffer III and Jennifer Neville Proceedings of the 5th International AAAI Conference on Weblogs and Social Media, ICWSM, 2011 [PDF] [BibTeX] [Extended Version] [MySQL 5.5 Database File] Using Touch to Localize Flexible Materials During Manipulation
Robert Platt, Jr., Frank Permenter, Joseph J. Pfeiffer IIIIEEE Transactions on Robotics, Special issue on a robotic sense of touch. June, 2011 [PDF] [BibTeX] Probabilistic Paths and Centrality in Time
Joseph J. Pfeiffer III and Jennifer Neville Proceedings of the 4th SNAKDD Workshop, KDD, 2010. [PDF] [BibTeX] [PPTX] Inferring handobject configuration directly from tactile data
Robert Platt, Jr., Frank Permenter, and Joseph J. Pfeiffer IIIElectronically published proceeding of the Mobile Manipulation Workshop, IEEE Conference on Robotics and Automation (ICRA), May 2010. [PDF] [BibTeX] A General Framework for Reconciling Multiple Weak segmentations of an Image
Soumya Ghosh, Joseph J. Pfeiffer III and J. Mulligan. IEEE Workshop on Applications of Computer Vision, 2009. [PDF] [BibTeX] Overcoming Uncertainty for WithinNetwork Relational Machine Learning
Joseph J. Pfeiffer III Ph.D. Thesis [PDF] [BibTeX] Codes
A collection of code for various models, generally written in Python for the simplest forms of representations. Several of these are done simply for my own practice/experience and I think they could be useful for someone else looking for example code. Some of these are implementations of other work and some from my own work above. Attributed Graph Models
Implementation of the above AGM model (WWW2014), using the simple FCL proposal distribution. Uses a simple 0/1 label value. Requires Python w/ matplotlib. The first implementation uses the Preferential Attachment model of Barabasi/Albert to create a degree distribution for AGM/FCL, while the second version learns/samples from an observed network.
[PAAGMFCL] [AGMFCL] Data Augmentation, Stochastic EM and EM
Implementation of the above ICDM2014 paper on Data Augmentation. Has some tests and comparisons between DA, Stochastic EM and EM for Naive Bayes and Logistic Regression. Implemented in C++ and contains a distribution of liblinear and eigen3. Please read and carry on any appropriate copyright notices for these works.
[Code] LogLinear models
A simple Python implementation for learning loglinear (maximum entropy) models. Just uses 0/1 feature/label values, and implemented for my own practice. Requires Python and scipy/numpy; I implemented both calling scipy's BFGS optimization, as well as my own GD method for fun.
[BFGS] [GD] 