WPI Worcester Polytechnic Institute

Computer Science Department
------------------------------------------

AIRG Topics - Spring 2016


Our group meets on Thursdays at 11:00 a.m.
  • Fall Semester: Beckett Conf. Rm.
  • Spring Semester: Beckett Conf. Rm.


Version: Wed 6 April 2016

Jan 14
No meeting

Jan 21
Xinyue Liu
Kernelized Matrix Factorization for Collaborative Filtering
    Matrix factorization (MF) methods have shown great promise in collaborative filtering (CF). Conventional MF methods usually assume that the correlated data is distributed on a linear hyperplane, which is not always the case. Kernel methods are used widely in SVMs to classify linearly non-separable data, as well as in PCA to discover the non-linear embeddings of data. In this paper, we present a novel method to kernelize matrix factorization for collaborative filtering, which is equivalent to performing the low-rank matrix factorization in a possibly much higher dimensional space that is implicitly defined by the kernel function. Inspired by the success of multiple kernel learning (MKL) methods, we also explore the approach of learning multiple kernels from the rating matrix to further improve the accuracy of prediction. Since the right choice of kernel is usually unknown, our proposed multiple kernel matrix factorization method helps to select effective kernel functions from the candidates. Through extensive experiments on real-world datasets, we show that our proposed method captures the nonlinear correlations among data, which results in improved prediction accuracy compared to the state-of-art CF models.

Jan 28
Carolina Ruiz
Deviation-based dynamic time warping for clustering human sleep
    In this talk, we propose two versions of a modified dynamic time warping approach for comparing discrete time series. This approach is motivated by the observation that the distribution of dynamic time warping paths between pairs of human sleep time series is concentrated around the path of constant slope. Both versions use a penalty term for the deviation between the warping path and the path of constant slope for a given pair of time series. In the first version, global weighted dynamic time warping, the penalty term is added as a postprocessing step after a standard dynamic time warping computation, yielding a modified similarity metric that can be used for time series clustering. The second version, stepwise deviation-based dynamic time warping, incorporates the penalty term into the dynamic programming optimization itself, yielding modified optimal warping paths, together with a similarity metric. Clustering experiments over synthetic data, as well as over human sleep data, show that the proposed methods yield significantly improved accuracy and generative log likelihood as compared with standard dynamic time warping.

Feb 4
tbd
tbd

Feb 11
tbd
tbd

Feb 18
Academic Advising Appointment Day
No AIRG Meeting

Feb 25
tbd
tbd

March 3
tbd
tbd

Mar 10
Ugrad & Grad student break

Mar 17
tbd
tbd

Mar 24
tbd
tbd

Mar 31
POSTPONED until 7th April
John Boaz Lee
Network Lasso: Clustering and Optimization in Large Graphs

Apr 7
John Boaz Lee
Network Lasso: Clustering and Optimization in Large Graphs
    Convex optimization is an essential tool for modern data analysis, as it provides a framework to formulate and solve many problems in machine learning and data mining. However, general convex optimization solvers do not scale well, and scalable solvers are often specialized to only work on a narrow class of problems. Therefore, there is a need for simple, scalable algorithms that can solve many common optimization problems. In this paper, we introduce the network lasso, a generalization of the group lasso to a network setting that allows for simultaneous clustering and optimization on graphs. We develop an algorithm based on the Alternating Direction Method of Multipliers (ADMM) to solve this problem in a distributed and scalable manner, which allows for guaranteed global convergence even on large graphs. We also examine a non-convex extension of this approach. We then demonstrate that many types of problems can be expressed in our framework. We focus on three in particular---binary classification, predicting housing prices, and event detection in time series data---comparing the network lasso to baseline approaches and showing that it is both a fast and accurate method of solving large optimization problems.

Apr 14
Yan Wang
Student modeling from different aspects
MS Thesis presentation
Advisor: Neil Heffernan
    With the wide usage of online tutoring systems, researchers become interested in mining data from logged files of these systems, so as to get better understanding of students. Varieties of aspects of students' learning have become focus of studies, such as modeling students. mastery status and affects. On the other hand, Randomized Controlled Trial (RCT), which is an unbiased method for getting insights of education, finds its way in Intelligent Tutoring System. Firstly, people are curious about what kind of settings would work better. Secondly, such a tutoring system, with lots of students and teachers using it, provides an opportunity for building a RCT infrastructure underlying the system. With the increasing interest in Data mining and RCT, the thesis focuses on these two aspects. In the first part, we focus on analyzing and mining data from ASSISTments, an online tutoring system run by a team in Worcester Polytechnic Institute. Through the data, we try to answer several questions from different aspects of students learning. The first question we try to answer is what matters more to student modeling, skill information or student information. The second question is whether it is necessary to model students. learning at different opportunity count. The third question is about the benefits of using partial credit, rather than binary credit as measurement of students. learning in RCT. The fourth question focuses on the amount that students spent Wheel Spinning in the tutoring system. The fifth questions studies the tradeoff between the mastery threshold and the time spent in the tutoring system. By answering the five questions, we both propose machine learning methodology that can be applied in educational data mining, and present findings from analyzing and mining the data. In the second part, we focused on RCTs within ASSISTments. Firstly, we looked at a pilot study of reassessment and relearning, which suggested a better system setting to improve students. robust learning. Secondly, we proposed the idea to build an infrastructure of learning within ASSISTments, which provides the opportunities to improve the whole educational environment.

Apr 21
Undergraduate Project Presentation Day
No AIRG Meeting

Apr 28
(reserved for possible MS thesis presentations)


[Feedback] [Search Our Web] [Help & Index]

[Return to the WPI Homepage] [Return to the CS Homepage]

AIRG Coordinator