Slides From Talks by Radford Neal

  • ``MCMC Training of Bayesian Neural Networks'', Machine Learning Advances and Applications, Fields Institute, 16 May 2022: slides, video.
  • ``MCMC for Hierachical Bayesian Models Using Non-reversible Langevin Methods'', virtual seminar at University of Oxford, 14 May 2020: slides, video.
  • ``Non-reversible Langevin Methods for Sampling Complex Distributions'', virtual seminar at McMaster University, 18 March 2020: PDF.
  • ``Non-reversibly updating a uniform [0,1] value for accept/reject decisions'', 2nd Symposium on Advances in Approximate Bayesian Inference, Vancouver, 8 December 2019: poster, paper.
  • ``Using Deterministic Maps when Sampling from Complex Distributions'', Evolution of Deep Learning Symposium (in honour of Geoffrey Hinton), 16 October 2019: PDF.
  • ``An Automatic Differentiation Extension for R, and its Implementation in pqR'', RIOT2019, Toulouse, 11 July 2019: PDF.
  • ``Automatic Differentiation for R'', Vector Research Symposium, 22 February 2019: PDF.
  • ``Recent and Planned Language Extensions in pqR'', RIOT2017, Brussels, 5 July 2017: PDF.
  • ``Advances in Memory Management and Symbol Lookup in pqR'', RIOT2017, Brussels, 5 July 2017: PDF.
  • ``Performance improvements and future language extensions in the pqR implementation of R'', Greater Toronto Area R User's Group, April 2016: PDF.
  • ``Reinforcement Learning with Randomization, Memory, and Prediction'', CRM - University of Ottawa Distinguished Lecture, April 2016: PDF.
  • ``Can Interpeting be as Fast as Byte Compiling? + Other Developments in pqR'', R Summit Conference, June 2015: PDF.
  • ``Learning to Randomize and Remmember in Partially-Observed Environments'', talk at the Fields Institute, Workshop on Big Data and Statistical Machine Learning, January 2015: PDF (video here).
  • ``Proposals for Extending the R language'', talk given at Directions in Statistical Computing (DSC 2014), Brixen / Bressanone, Italy, June 2014: PDF.
  • ``Speed Improvements in pqR: Current Status and Future Plans'', talk given at Directions in Statistical Computing (DSC 2014), Brixen / Bressanone, Italy, June 2014: PDF.
  • ``Speeding up R with Multithreading, Task Merging, and Other Techniques'', talk given at the University of Guelph, Dept. of Mathematics & Statistics, 29 November 2013: PDF.
  • ``Probability and Anthropic Reasoning in Small, Large, and Infinite Universes'', Conference on Challenges for Early Universe Cosmology, Perimeter Institute, July 2011: PDF of slides (but note that much of the talk was on the blackboard).
  • ``New Monte Carlo Methods Based on Hamiltonian Dynamics'', MaxEnt 2011, July 2011: PDF.
  • ``MCMC Using Ensembles of States with Application to Gaussian Process Regression'', talk given at the University of Toronto Dept. of Economics (2010-10-22) and Dept. of Computer Science (2010-11-15): Postscript, PDF.
  • ``Nuisance Parameters and Other Issues in Searching for Signals in High-Energy Physics Experiments'', Talk (remotely) at the PHYSTAT-LHC Workshop , June 2007: Postscript, PDF.
  • ``Short-Cut MCMC: An Alternative to Adaptation'', Talk at the Third Workshop on Monte Carlo Methods, May 2007: Postscript, PDF.
  • ``Constructing Efficient MCMC Methods Using Temporary Mapping and Caching'', Talk at Columbia University, December 2006: Postscript, PDF.
  • ``Estimation of Failure Probabilities Using Linked Importance Sampling'', Workshop on Nonlinearlity and Randomness in Complex Systems, SUNY at Buffalo, April 2006: Postscript (0.4 MBytes), PDF (4.7 MBytes).
  • ``Hamiltonian Importance Sampling'', BIRS workshop on Mathematical Issues in Molecular Dynamics, June 2005: Postscript, PDF.
  • ``Creating Non-Gaussian Processes from Gaussian Processes by the Log-Sum-Exp Approach'', talk to the U of T machine learning group, February 2005: Postscript, PDF.
  • NIPS*2004 tutorial on ``Bayesian Methods for Machine Learning'': Postscript, PDF.
  • ``A New Proof of Peksun's Theorem Regarding the Asymptotic Variance of MCMC Estimators'', poster at ISBA 2004, Vina del Mar, Chile, May 2004: postscript, pdf.
  • ``Classification for High Dimensional Problems Using Bayesian Neural Networks and Dirichlet Diffusion Trees'', NIPS*2003 Feature Selection Workshop, Whistler, British Columbia, December 2003 (describing the winning entry): pdf.
  • ``Markov chain Monte Carlo computations for Dirichlet diffusion trees'', NTOC 2001, Kyoto, December 2001: postscript, pdf.
  • ``Survival analysis using a Bayesian neural network'', Joint Statistical Meetings, Atlanta, 2001: postscript, pdf.
  • ``Monte Carlo decoding of LDPC codes'', ICTP Workshop on Statistical Physics and Capacity-Approaching Codes, May 2001: postscript, pdf.
  • ``Circularly-coupled Markov chain sampling'', April 2000: postscript, pdf. See also the earlier technical report.
  • ``Markov chain sampling using Hamiltonian dynamics'', Joint Statistical Meetings, Baltimore, 1999: postscript, pdf.
  • ``Faster encoding for low-density parity check codes using sparse matrix methods'', IMA workshop on Codes, Systems and Graphical Models, Minneapolis, 1999: postscript, pdf.
  • ``Tutorial on exact sampling methods'', given 26 October 1998 during the workshop on Monte Carlo methods at the Fields Institute: postscript, pdf.
  • ``Improving Markov chain sampling by suppressing random walks'', AMS-IMS-SIAM Joint Workshop on Stochastic Inference, Monte Carlo, and Empirical Methods, July 1996: postscript, pdf.
    Back to Radford Neal's home page