Computer Vision Reading Group (CVRG)

Spring 2013
Day/time: Thursday 4pm-5pm
Location: CS 4310
Mailing list: sign up here
Organizer: Jia Xu


In this reading group, we will discuss recent works on Computer Vision, as well as interesting topics on Machine Learning, Computer Graphics and Optimization. Every other week a different volunteer will present a chosen research paper relevant to computer vision and lead an informal discussion. "Off" weeks will be reserved for people interested in presenting and discussing their own research. If you want to lead a discussion on some paper(s) or present your own research at CVRG, please send me an email.


Week 1 (02/07/2013): Brandon Smith

  • Ce Liu, Jenny Yuen, and Antonio Torralba. Nonparametric Scene Parsing via Label Transfer, PAMI 2011.

  • Week 2 (02/21/2013): Vamsi Ithapu

  • Nitish Srivastava and Ruslan Salakhutdinov. Multimodal Learning with Deep Boltzmann Machines. Neural Information Processing Systems (NIPS 26), 2012.

  • Week 3 (03/07/2013): Deepti Pachauri

  • Richard Roberts, Sudipta N. Sinha, Richard Szeliski, Drew Steedly. Structure from motion for scenes with large duplicate structures. CVPR 2011.

  • Week 4 (03/21/2013): Moo K. Chung, Associate Professor, Biostatistics and Medical Informatics.

  • Title: Exploiting hidden persistent homological structures in compressed sensing and sparse likelihood and its application to graphs and networks.
  • Abstract: We will explicitly show how to identify persistent homology in compressed sensing and sparse likelihood problems related to graphs/networks and completely bypass time consuming numerical optimization. The following paper will be discussed in detail:

  • Week 5 (04/04/2013): Jerry Zhu, Associate Professor, Computer Sciences.

  • Title: Introduction to Persistent Homology
  • Time: April 4, 3:30pm-4:30pm, CS4310 ( note we will be half an hour earlier )
  • Abstract: Persistent homology is a rapidly growing branch of topology, gaining increasing interest in the machine learning community. The 0-th order homology groups correspond to clusters, which are the bread-and-butter of modern data analysis. The 1st order homology groups are "holes," as in the center of a donut; The 2nd order homology groups are "voids," as the inside of a balloon; and so on. These seemingly exotic higher-order mathematical structures may provide valuable invariant data representations that complement current feature-based representations. This talk will be a gentle introduction to persistent homology accessible to all computer scientists.

  • Week 6 (04/25/2013): Jie Liu

  • Title: Graphical-model Based Multiple Testing under Dependence, with Applications to Genome-wide Association Studies.
  • Time/place: April 25, 4pm-5pm, CS4310 
  • Abstract: Large-scale multiple testing tasks often exhibit dependence, and leveraging the dependence between individual tests is still one challenging and important problem in statistics. With recent advances in graphical models, it is feasible to use them to perform multiple testing under dependence. We propose a multiple testing procedure which is based on a Markov-random-field-coupled mixture model. The ground truth of hypotheses is represented by a latent binary Markov random field, and the observed test statistics appear as the coupled mixture variables. The parameters in our model can be automatically learned by a novel EM algorithm. We use an MCMC algorithm to infer the posterior probability that each hypothesis is null, and the false discovery rate can be controlled accordingly. Simulations show that the numerical performance of multiple testing can be improved substantially by using our procedure. We apply the procedure to a real-world genome-wide association study on breast cancer, and we identify several SNPs with strong association evidence.

  • Week 7 (05/16/2013): Jia Xu

  • Title: Incorporating Topological Constraints within Interactive Segmentation and Contour Completion via Discrete Calculus.
  • Time/place: May 16, 4pm-5pm, CS4310 
  • Abstract: We study the problem of interactive segmentation and contour completion for multiple objects. The form of constraints our model incorporates are those coming from user scribbles (interior or exterior constraints) as well as information regarding the topology of the 2-D space after partitioning (number of closed contours desired). We discuss how concepts from discrete calculus and a simple identity using the Euler characteristic of a planar graph can be utilized to derive a practical algorithm for this problem. We also present specialized branch and bound methods for the case of single contour completion under such constraints. On an extensive dataset of ~1000 images, our experiments suggest that a small amount of side knowledge can give strong improvements over fully unsupervised contour completion methods. We show that by interpreting user indications topologically, user effort is substantially reduced. More details can be found at