Algorithms & Computationally Intensive Inference seminars
2020/2021: The seminars will happen on Microsoft Teams on Fridays 1pm UK time (or occasionally at different times).
If you are not affiliated with Warwick and wish to attend our seminars please register here.
If you would like to speak, or you want to be included in any emails, please contact one of the organisers.
Current Organisers: Alice Corbella, Massimiliano Tamborrino, Jure Vogrinc
Website URL: www.warwick.ac.uk/compstat
Mailing List Sign-Up: http://mailman1.csv.warwick.ac.uk/mailman/listinfo/algorithmseminar
Mailing List: algorithmseminar@listserv.csv.warwick.ac.uk (NB - only approved members can post)
Microsoft Teams Link (username/password same as for eduroam) available here
Next Talk: Friday June 25th, 1pm UK time
Sophie Langer: Deep Learning meets statistics: Improving neural networks with statistical theory
Abstract: In this talk we analyse the L2 error of neural network regression estimates with one hidden layer. Under the assumption that the Fourier transform of the regression function decays suitably fast, we show that an estimate, where all initial weights are chosen according to proper uniform distributions and where the weights are learned by gradient descent, achieves a rate of convergence of 1/ n (up to a logarithmic factor). Our statistical analysis implies that the key aspect behind this result is the proper choice of the initial inner weights and the adjustment of the outer weights via gradient descent. This indicates that we can also simply use linear least squares to choose the outer weights. We prove a corresponding theoretical result and compare our new linear least squares neural network estimate with standard neural network estimates via simulated data. Our simulations show that our theoretical considerations lead to an estimate with an improved performance. Hence the development of statistical theory can indeed improve neural network estimates.
2020/21 Term 3:
| Date | Speaker | Title | Abstract | Slides | Video |
| Week 1: 30/4 | Jure Vogrinc | Counterexamples for optimal scaling of Metropolis-Hastings chains with rough target densities | Abstract | Slides | Video |
| Week 2: 7/5 | Gilles Louppe | The frontiers of simulation-based inference | Abstract | Video | |
| Week 3: 14/5 | Andrea Bertazzi | Euler approximations for Piecewise deterministic Markov processes | Abstract | Video | |
| Week 4: 21/5 | Juan Kuntz Nussio | Product-form estimators: exploiting independence to scale up Monte Carlo | Abstract | ||
| Week 5: 28/5 | Dootika Vats | The new BFFs in MCMC: Barker and Bernoulli | Abstract | Video | |
| Week 6: 4/6 | Anthony Lee | A general perspective on the Metropolis-Hastings kernel: incorporating stopping times in proposals. | Abstract | Slides | Video |
| Week 7: 11/6 | Timothée Stumpf-Fétizon | Exact Bayesian Inference for Markov Switching Diffusions | Abstract | Slides | Video |
| Week 8: 18/6 | Helen Ogden |
Information criteria for model choice in finite mixture models |
Abstract | ||
| Week 9: 25/6 | Sophie Langer | Deep Learning meets statistics: Improving neural networks with statistical theory | Abstract | Slides | Video |
| Week 10: 2/7 | CANCELLED | DUE TO ISBA MEETING |
2020/21 Term 2:
| Date | Speaker | Title | Abstract | Slides | Video |
| Week 1: 15/1 (4pm UK time) | Jeremias Knoblauch | Postponed | Abstract | ||
| Week 2: 22/1 (4pm UK time) | Jeffrey Rosenthal | MCMC Confidence Intervals and Biases Without CLTs | Abstract | Slides | Video |
| Week 3: 29/1 | Paul Dobson | Subgeometric hypocoercivity for piecewise-deterministic Markov process Monte Carlo methods | Abstract | Slides | Video |
| Week 4: 5/2 (4:30 UK time) | Daniel Jerison | MCMC convergence bounds for reversible chains | Abstract | Slides | Video |
| Week 5: 12/2 | Jaromir Sant | Inference of natural selection from allele frequency time series data using exact simulation techniques | Abstract | Slides | Video |
| Week 6: 19/2 | Xenia Miscouridou |
Modelling sparsity, heterogeneity, reciprocity and community structure in temporal interaction data |
Abstract | Video | |
| Week 7: 26/2 | Nianqiao (Phyllis) Ju | Sequential Monte Carlo algorithms for agent-based models of disease transmission | Abstract | Slides | Video |
| Week 8: 5/3 | Adeline Samson |
Computational statistics for neuronal mathematical models
|
Abstract | Slides | Video |
| Week 9: 12/3 (4 pm UK time) | Ari Stern | Structure-preserving numerical integrators: classic methods, new perspectives | Abstract | Video | |
| Week 10: 19/3 | Jeremias Knoblauch | Optimization-centric Generalizations of Bayesian Inference | Abstract | Slides | Video |
This is the list of confirmed speakers, which will be continuously updated.
| Date | Speaker | Title | Abstract | Slides | Video |
| Week -2: 18/09 (12 UK time) | Clara Grazian | Approximate Bayesian analysis of (un)conditional copulas | Abstract | Slides | Video |
| Week -1: 25/09 | Joe Meagher | Bayesian Ancestral Reconstruction for Bat Echolocation | Abstract | Slides | Video |
| Week0: 02/10 | Michael Choi | On the convergence of an improved and adaptive kinetic simulated annealing | Abstract | Slides | Video |
| Week 1: 09/10 | cancelled (OxWaSP workshop) | ||||
| Week 2: 16/10 (2pm UK time) | Liangliang Wang | Sequential Monte Carlo for estimating parameters of differential equations | Abstract | Slides | Video |
| Week 3: 23/10 | Sebastian Vollmer |
Part 1: Risk prediction and Risk prediction Part 2: Machine Learning in Julia and benchmarking results on predictive fairness. |
Abstract | Video | |
| Week 4: 30/10 (4pm UK time) | Philippe Gagnon | Lifted samplers for partially ordered discrete state-spaces | Abstract | Video | |
| Week 5: 6/11 | Richard Everitt |
Rare event ABC-SMC^2 |
Abstract | Slides | Video |
| Week 6: 13/11 | Letizia Angeli | Interacting Particle Systems Approximations of Feynman-Kac Formulae in continuous time | Abstract | Slides | Video |
| Week 7: 20/11 | Yan Qu | Exact Simulation of Self-Excited Point Process with Levy Driven OU | Abstract | Video | |
| Week 8: 27/11 (4pm UK time) |
Caroline Colijn |
COVID-19 data sources and their challenges for modelling and estimation
|
Abstract | Video | |
| Week 9: 4/12 | Suzie Brown | Asymptotic genealogies of sequential Monte Carlo algorithms | Abstract | Video | |
| Week 10: 11/12 | cancelled |
Previous Years:
Some key phrases:
- Sampling and inference for diffusions
- Exact algorithms
- Intractable likelihood
- Pseudo-marginal algorithms
- Particle filters
- Importance sampling
- MCMC
- Adaptive MCMC
- Perfect simulation
- Markov chains
- Random structures
- Randomised algorithms