Roland Memisevic University of Frankfurt I received the PhD in machine learning from the University of Toronto in 2008 working with Geoffrey Hinton. I am currently an assistant professor (German: "Junior professor") at the Department of Computer Science, University of Frankfurt, Germany. To contact me write to roland[at]cs[dot]toronto[dot]edu or ro[at]cs[dot]uni-frankfurt[dot]de. Brief bio.

• I will give a tutorial on spatio-temporal feature learning at CVPR 2012.


• Complex cells and energy models perform joint diagonalization (allowing them to learn dynamic pinwheels).
  
  
• Multi-view feature learning extends dictionary learning to stereo and other multi-observation settings.
  (Think sparse coding on natural images always gives you Gabors? Think twice!)
  
  

Publications

• 2012 Memisevic, R.
  On multi-view feature learning.
  International Conference on Machine Learning ICML 2012
  [preprint]
  
  
• 2012 Memisevic, R., Sigal, L., Fleet, D.
  Shared Kernel Information Embedding for Discriminative Inference.
  IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI)
  [preprint] © IEEE [code]
  
  
• 2011 Memisevic, R.
  Gradient-based learning of higher-order image features.
  International Conference on Computer Vision (ICCV 2011).
  [pdf] [bibtex] © IEEE [Website]
  
  
• 2011 Susskind, J., Memisevic, R., Hinton, G., Pollefeys, M.
  Modeling the joint density of two images under a variety of transformations.
  Computer Vision and Pattern Recognition (CVPR 2011).
  [pdf] [bibtex] [code] © IEEE
  
  
• 2011 Memisevic, R.
  Learning to relate images: Mapping units, complex cells and simultaneous eigenspaces.
  arXiv:1110.0107v1 [pdf] [bibtex]
  
  
• 2011 Memisevic, R., Conrad, C.
  Depth as a latent variable.
  Frontiers in Neuroscience Conference Abstract: BC11
  [link]
  
  
• 2011 Memisevic, R.
  On spatio-temporal sparse coding: Analysis and an algorithm.
  NIPS workshop on Deep Learning and Unsupervised Feature Learning 2011
  
  
• 2011 Memisevic, R., Conrad, C.
  Stereopsis via Deep Learning.
  NIPS workshop on Deep Learning and Unsupervised Feature Learning 2011
  
  
• 2010 Memisevic, R., Hinton, G.
  Learning to Represent Spatial Transformations with Factored Higher-Order Boltzmann Machines.
  June 2010 edition, Vol. 22, No. 6: 1473-1492, Journal Neural Computation.
  [pdf], [bibtex], [Website]
  related 2009 technical report: [pdf]
  
  
• 2010 Memisevic, R., Zach, C., Hinton, G., Pollefeys, M.
  Gated Softmax Classification.
  Neural Information Processing Systems (NIPS 2010).
  [pdf], [bibtex], [Website].
  
  
• 2009 Sigal, L., Memisevic, R., Fleet, D.
  Shared kernel information embedding for discriminative inference.
  Computer Vision and Pattern Recognition (CVPR 2009 © IEEE).
  [pdf]
  
  
• 2008 Memisevic, R.
  Non-linear latent factor models for revealing structure in high-dimensional data.
  PhD thesis, University of Toronto 2008.
  [pdf], [bibtex]
  
  
• 2007 Memisevic, R. and Hinton, G. E.
  Unsupervised learning of image transformations.
  Computer Vision and Pattern Recognition (CVPR 2007 © IEEE).
  [pdf], [bibtex]
  (Related Technical Report: [pdf])
  
  
• 2007 Samulowitz, H. and Memisevic, R.
  Learning to solve QBF.
  Twenty-Second Conference on Artificial Intelligence (AAAI 2007).
  [pdf][bibtex]
  
  
• 2006 Memisevic, R.
  An introduction to structured discriminative learning. Technical report.
  University of Toronto, 2006. [ps],[pdf]
  
  
• 2006 Memisevic, R.
  Dual optimization of conditional probability models.
  NIPS Workshop on Kernel Methods and Structured Domains
  (Technical report, University of Toronto, 2006: [ps],[pdf])
  
  
• 2006 Memisevic, R.
  Kernel Information Embeddings.
  23rd International Conference on Machine Learning (ICML 2006).
  [pdf],[bibtex], [Python code], [GPU version].
  
  
• 2006 Memisevic, R.
  Propagating Errors and Beliefs for Large Scale Nonlinear Structure Prediction.
  North East Student Colloquium on Artificial Intelligence. Ithaca, NY.
  
  
• 2005 Meinicke, P., Klanke, S., Memisevic, R., Ritter, H.
  Principal Surfaces from Unsupervised Kernel Regression.
  IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI)
  Vol. 27, no. 9, pp. 1379-1391. [bibtex]
  
  
• 2005 Memisevic, R. and Hinton, G. E.
  Improving dimensionality reduction with spectral gradient descent.
  Journal Neural Networks.
  18, pp 702-710.
  [online version],[bibtex]
  
  
• 2005 Memisevic, R. and Hinton, G. E.
  Embedding via clustering: Using spectral information to guide dimensionality reduction.
  International Joint Conference on Neural Networks 2005.
  
  
• 2004 Memisevic, R. and Hinton, G. E.
  Multiple Relational Embedding.
  Advances in Neural Information Processing Systems (NIPS 2004).
  [ps.gz],[pdf],[bibtex]
  
  
• 2003 Memisevic, R.
  Unsupervised Kernel Regression for Nonlinear Dimensionality Reduction. [pdf]
  Master's Thesis, Bielefeld University, 2003.
  The thesis describes a framework, based on kernel density estimation, that reconciles two common, but previously
  separate, views of dimensionality reduction:
  Spectral embedding (such as Laplacian Eigenmaps, ISOMAP, LLE) which allows for efficent, global optimization,
  vs. generative models (aka. latent variable models: Factor Analysis, GTM, Principal Curves, etc.), that provide
  proper density estimates and out-of-sample generalizations but are inefficient and based on non-global optimization.
  Check out this talk (pdf) for the basic ideas.
  

Professional activities

• Senior PC, PC or reviewer for CVPR 2004, 2005, 2010, 2011 (PC), 2012 (PC); NIPS 2006, 2007, 2008, 2009, 2011; ICCV 2010, 2011 (PC); ECCV 2010; DAGM 2011 (Senior PC), 2012 (Senior PC); ICML 2006, 2008 (PC), 2012 (PC).
• Reviewer for IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), Journal of Machine Learning Research (JMLR), Journal Image and Vision Computing (IMAVIS), Journal Constraint, Journal Machine Learning, IEEE Transactions on Image Processing, International Journal of Adaptive Control and Signal Processing.
• Grant reviewer, Swiss National Science Foundation.
• Tutorial at CVPR 2012 and DAGM 2011 on higher-order feature learning and on spatio-temporal learning [abstract] [website].
  
  

Teaching

• Winter 2011/12: Instructor, Machine Learning, University of Frankfurt.
  
• Spring 2011: Instructor, Digital Image Processing (in German), University of Frankfurt.
  
• Before 2011: Teaching Assistant at the University of Toronto: Introduction to Neural Networks and Machine Learning (CSC321, 2006 and 2007), Software Design (CSC207, 2006), Machine Learning (CSC2515, 2005 and 2006).

Some slides from selected talks

• May 2011 Memisevic, R.
  Learning to relate images
  Max Planck Institute for Biological Cybernetics, Tuebingen.
• May 2010 Memisevic, R.
  How to train a mixture of 100.000.000.000.000.000 classifiers
  University of Toronto Machine Learning Seminar
• March 2007 Memisevic, R., Hinton, G. E.
  Unsupervised learning of image transformations
  University of Toronto Machine Learning Seminar
• June 2006 Memisevic, R.
  Kernel Information Embeddings
  ICML 2006
• Dec 2005 Memisevic, R.
  CRFs in the dual.
  NIPS Workshop on Kernel Methods and Structured Domains.
• Dec 2005 Memisevic, R., Srebro, N., Roweis, S.
  Exploring the regularization path for adaptive Gaussian kernel SVMs.
  NIPS Workshop on the Accuracy-Regularization Frontier.

Miscellaneous

• CSC 321: Slides for the tutorials on Matlab for novices, Matlab for machine learning and Probabilities for machine learning
• A Python module implementing hidden Markov models: hmm.py
  Supports discrete or Gaussian observations and training on a single or on multiple observation sequences.
• minimize.py is a Python adaptation of Carl Rasmussen's matlab function minimize.m
• Slides and a few code-snippets for a ten-minute mini-tutorial on switching from Matlab to Python.

Contact information