.

Omar Rivasplata
Centre for AI Fundamentals Department of Computer Science University of Manchester

Noteworthy News ‖ CV ‖ Email: try or ‖ Math Genealogy ‖ Follow:
Statistical Learning Theory. Machine Learning. AI. Mathematics. Probability and Statistics.

I certify AI predictions with high confidence.
My research is on machine learning, theory and practice. That said, I tend to have a broad interest in various aspects of the mathematical and statistical foundations of machine learning and AI in general. In the limit of tending to praxis, I work on designing strategies to train and certify machine learning models. Previously, I have led or been a contributor to projects on offline reinforcement learning, generative models, PACBayes bounds for deep learning and kernel classifiers, among other.
The field of machine learning research is fascinating! One of the things I enjoy most about my work being the confluence of maths and stats, and computer experiments with collaborators, to answer questions about the optimisation and certification of machine learning models. Besides statistical learning I am interested also in online learning and reinforcement learning. Of course I am interested in deep learning, which is quite popular these days. Optimisation is a pervasive theme across machine learning theory and practice, though it comes up in such a variety of flavours and colours that it isn't boring. It reminds of the least action principle of Maupertuis, saying that "everything happens as if some quantity was to be made as small as possible." (This principle has lead the optimists to believe that we live in the best possible world.) But just optimisation doesn't quite do it for machine learning... to really be talking about learning one has to pay attention to learner's performance beyond the data used for training; so certification is very important!
These are my affiliations with learned sociaties:

For a couple of years and a half I was a Senior Research Fellow in the Department of Statistical Science at UCL, where I created and led the research group DELTA. Mine was one of two departments of the upcoming Institute for Mathematical and Statistical Sciences, the other one being the Department of Mathematics.
My dissertation PACBayesian Computation was the result of my research studies in statistical learning at the Department of Computer Science, University College London, which were sponsored by DeepMind. In parallel with my research studies at UCL, I had a fixedterm position with DeepMind for three years.
Outstanding ML/AI people with whom I have worked at UCL Computer Science include David Barber, Marc Deisenroth, Mark Herbster, María PérezOrtiz, and John ShaweTaylor.
Outstanding ML/AI people with whom I have worked at DeepMind include Marcus Hutter, Laurent Orseau, Ilja Kuzborskij, Csaba Szepesvári, András György from my own team; and many others from friend teams including Amal RannenTriki, Razvan Pascanu, Agnieszka GrabskaBarwińska, Thore Graepel, Arnaud Doucet, Benjamin Van Roy, and Geoffrey Irving. 
My journey into ML research started at the Department of Computing Science, University of Alberta, where I spent a year and two months. During this time I started building my mental model of the ML field and finetuning my hyperparameters. My host was Rich Sutton in theory, but in practice I was working with Csaba Szepesvári.

 Tighter risk certificates for (probabilistic) neural networks.
UCL Centre for AI.
Slides
Video
 Statistical Learning Theory: A Hitchhiker's Guide.
NeurIPS 2018 Tutorial. (with J. ShaweTaylor)
Slides
Video
Conference & Journal Papers 
 P. Blomstedt, D. Mesquita, O. Rivasplata et al.
Metaanalysis of Bayesian Analyses.
To appear in Bayesian Analysis, 2024.
 S.D. Mbacke and O. Rivasplata,
A Note on the Convergence of Denoising Diffusion Probabilistic Models.
TMLR 2024.
PDF
 B. Rodríguez Gálvez, O. Rivasplata, R. Thobaben, M. Skoglund,
A note on generalization bounds for losses with finite moments.
ISIT 2024.
arXiv PDF
 I. Kuzborskij, Cs. Szepesvári, O. Rivasplata, A. RannenTriki, R. Pascanu,
On the Role of Optimization in Double Descent: A Least Squares Study.
NeurIPS 2021.
arXiv PDF
 M. PérezOrtiz, O. Rivasplata, J. ShaweTaylor, Cs. Szepesvári,
Tighter risk certificates for neural networks.
JMLR, 22, 227 (2021), 140.
PDF /
revised PDF /
published PDF
 L. Orseau, M. Hutter, O. Rivasplata,
Logarithmic pruning is all you need.
NeurIPS 2020 .
PDF
 O. Rivasplata, I. Kuzborskij, Cs. Szepesvári, J. ShaweTaylor,
PACBayes analysis beyond the usual bounds.
NeurIPS 2020.
PDF
 O. Rivasplata, E. ParradoHernández, J. ShaweTaylor, S. Sun, Cs. Szepesvári,
PACBayes bounds for stable algorithms with instancedependent priors.
NeurIPS 2018.
PDF
 A.E. Litvak, O. Rivasplata,
Smallest singular value of sparse random matrices.
Studia Math., 212, 3 (2012), 195218.
PDF
 O. Rivasplata, J. Rychtar, B. Schmuland,
Reversibility for diffusions via quasiinvariance.
Acta Univ. Carolin. Math. Phys., 48, 1 (2007), 310.
PDF
 O. Rivasplata, J. Rychtar, C. Sykes,
Evolutionary games in finite populations.
Pro Mathematica, 20, 39/40 (2006), 147164.
PDF
 O. Rivasplata, B. Schmuland,
Invariant and reversible measures for random walks on Z.
Pro Mathematica, 19, 37/38 (2005), 117124.
PDF
 O. Rivasplata,
A note on a confidence bound of Kuzborskij and Szepesvári.
(2021)
PDF
 O. Rivasplata,
Subgaussian random variables: An expository note.
(2012)
PDF
 O. Rivasplata, V. Tankasali, Cs. Szepesvári,
PACBayes with Backprop.
(2019)
PDF
Probability Links (probably accessible) 
Stats Links (most significant) 
Canadian Links 
Peruvian Links 
My birth town is Trujillo, the marinera dance town. 
Sometimes people ask me about Machu Picchu, it's a great place to see. 
They ask me less about Arequipa though it is also a great place to visit. 
Last link, in case you care to know, is about Pisco. 

