Ole Winther

Professor in High dimensional biological data analysis/Machine learning at Section for Computational and RNA Biology, Department of Biology, University of Copenhagen and professor in Data science and complexity at Section for Cognitive Systems, DTU Compute, Technical University of Denmark.

Research interests:

I'm CRO (chief research officer) and co-founder of raffle.ai, where we use modern NLP for enterprise search.
I'm CTO and co-founder of FindZebra, a search engine for rare diseases.
I'm Head of the ELLIS Unit Copenhagen and a co-PI of the Machine Learning for Life Science (MLLS) Center.

E-mail: ole.winther@bio.ku.dk, olwi@dtu.dk and ole.winther@gmail.com

Current research group: Former group members

Research overview - tl; dr;

  • Use modern deep learning methodology for predictive and generative tasks in bioinformatics, NLP and science.
  • Develop inference algorithms and architectures for probabilistic models such a variational autoencoders.


Biological sequence analysis for proteins such as signal peptides (SignalP), subcellular localisation (DeepLoc) and transmembrane topology (DeepTMHMM). Understanding regulation on single cell gene expression, RNA and DNA level.

Latent variable models

Long lasting interest in inference algorithms and architectures for latent variable models:

Natural language processing

Search for health informatics, understanding large language models in the medical context and developing algorithms for end-to-end learning of joint retrieval and generation models.

AI for science

Making datasets for training models that can act as fast surrogates for physical models such as density functional theory (DFT).

Selected publicatons - never up to date ;-)

scVAE: variational auto-encoders for single-cell gene expression data

Christopher Heje Grønbech, Maximillian Fornitz Vording, Pascal N Timshel, Casper Kaae Sønderby, Tune H Pers, Ole Winther
Bioinformatics, 2020
Link Optimal Variance Control of the Score Function Gradient Estimator for Importance Weighted Bounds

This paper introduces novel results for the score function gradient estimator of the importance weighted variational bound (IWAE). We prove that in the limit of large K (number of importance samples) one can choose the control variate such that the Signal-to-Noise ratio (SNR) of the estimator grows as √K. This is in contrast to the standard pathwise gradient estimator where the SNR decreases as 1/√K. Based on our theoretical findings we develop a novel control variate that extends on VIMCO. Empirically, for the training of both continuous and discrete generative models, the proposed method yields superior variance reduction, resulting in an SNR for IWAE that increases with K without relying on the reparameterization trick. The novel estimator is competitive with state-of-the-art reparameterization-free gradient estimators such as Reweighted Wake-Sleep (RWS) and the thermodynamic variational objective (TVO) when training generative models.

Valentin Liévin, Andrea Dittadi, Anders Christensen, Ole Winther
NeurIPS, 2020
Link BIVA: A Very Deep Hierarchy of Latent Variables for Generative Modeling

With the introduction of the variational autoencoder (VAE), probabilistic latent variable models have received renewed attention as powerful generative models. However, their performance in terms of test likelihood and quality of generated samples has been surpassed by autoregressive models without stochastic units. Furthermore, flow-based models have recently been shown to be an attractive alternative that scales well to high-dimensional data. In this paper we close the performance gap by constructing VAE models that can effectively utilize a deep hierarchy of stochastic variables and model complex covariance structures. We introduce the Bidirectional-Inference Variational Autoencoder (BIVA), characterized by a skip-connected generative model and an inference network formed by a bidirectional stochastic inference path.

Lars Maaløe, Marco Fraccaro, Valentin Liévin, Ole Winther
NeurIPS, 2019
DeepLoc: prediction of protein subcellular localization using deep learning

Jose Juan Almagro Armenteros, Casper Kaae Sønderby, Søren Kaae Sønderby, Henrik Nielsen, Ole Winther
Bioinformatics, 2017
Link Ladder Variational Autoencoders

Variational Autoencoders are powerful models for unsupervised learning. However deep models with several layers of dependent stochastic variables are difficult to train which limits the improvements obtained using these highly expressive models. We propose a new inference model, the Ladder Variational Autoencoder, that recursively corrects the generative distribution by a data dependent approximate likelihood in a process resembling the recently proposed Ladder Network. We show that this model provides state of the art predictive log-likelihood and tighter log-likelihood lower bound compared to the purely bottom-up inference in layered Variational Autoencoders and other generative models.

Casper Kaae Sønderby, Tapani Raiko, Lars Maaløe, Søren Kaae Sønderby, Ole Winther
NeurIPS, 2016
Link Sequential Neural Models with Stochastic Layers

How can we efficiently propagate uncertainty in a latent state representation with recurrent neural networks? This paper introduces stochastic recurrent neural networks which glue a deterministic recurrent neural network and a state space model together to form a stochastic and sequential neural generative model. The clear separation of deterministic and stochastic layers allows a structured variational inference network to track the factorization of the model's posterior distribution. By retaining both the nonlinear recursive structure of a recurrent neural network and averaging over the uncertainty in a latent path, like a state space model, we improve the state of the art results on the Blizzard and TIMIT speech modeling data sets by a large margin, while achieving comparable performances to competing methods on polyphonic music modeling.

Marco Fraccaro, Søren Kaae Sønderby, Ulrich Paquet, Ole Winther
NeurIPS, 2016, Oral
Link Auxiliary Deep Generative Models

Deep generative models parameterized by neural networks have recently achieved state-of-the-art performance in unsupervised and semi-supervised learning. We extend deep generative models with auxiliary variables which improves the variational approximation. The auxiliary variables leave the generative model unchanged but make the variational distribution more expressive. Inspired by the structure of the auxiliary variable we also propose a model with two stochastic layers and skip connections. Our findings suggest that more expressive and properly specified deep generative models converge faster with better results. We show state-of-the-art performance within semi-supervised learning on MNIST, SVHN and NORB datasets.

Lars Maaløe, Casper Kaae Sønderby, Søren Kaae Sønderby, Ole Winther
ICML, 2016
Link Autoencoding beyond pixels using a learned similarity metric

We present an autoencoder that leverages learned representations to better measure similarities in data space. By combining a variational autoencoder (VAE) with a generative adversarial network (GAN) we can use learned feature representations in the GAN discriminator as basis for the VAE reconstruction objective. Thereby, we replace element-wise errors with feature-wise errors to better capture the data distribution while offering invariance towards eg translation.

Anders Boesen Lindbo Larsen, Søren Kaae Sønderby, Hugo Larochelle, Ole Winther
ICML, 2016
Perturbative Corrections for Approximate Inference in Gaussian Latent Variable Models

Expectation Propagation (EP) provides a framework for approximate inference. When the model under consideration is over a latent Gaussian field, with the approximation being Gaussian, we show how these approximations can systematically be corrected. A perturbative expansion is made of the exact but intractable correction, and can be applied to the model's partition function and other moments of interest. The correction is expressed over the higher-order cumulants which are neglected by EP's local matching of moments. Through the expansion, we see that EP is correct to first order. By considering higher orders, corrections of increasing polynomial complexity can be applied to the approximation. The second order provides a correction in quadratic time, which we apply to an array of Gaussian process and Ising models. The corrections generalize to arbitrarily complex approximating families, which we illustrate on tree- structured Ising model approximations. Furthermore, they provide a polynomial-time assessment of the approximation error. We also provide both theoretical and practical insights on the exactness of the EP solution.

Manfred Opper, Ulrich Paquet, Ole Winther
JMLR, 2013