Martin Jankowiak
YOU?
Author Swipe
View article: CRISPR-CLEAR: Nucleotide-Resolution Mapping of Regulatory Elements via Allelic Readout of Tiled Base Editing
CRISPR-CLEAR: Nucleotide-Resolution Mapping of Regulatory Elements via Allelic Readout of Tiled Base Editing Open
CRISPR tiling screens have advanced the identification and characterization of regulatory sequences but are limited by low resolution arising from the indirect readout of editing via guide RNA sequencing. This study introduces CRISPR-CLEAR…
View article: Joint genotypic and phenotypic outcome modeling improves base editing variant effect quantification
Joint genotypic and phenotypic outcome modeling improves base editing variant effect quantification Open
View article: Reparameterized Variational Rejection Sampling
Reparameterized Variational Rejection Sampling Open
Traditional approaches to variational inference rely on parametric families of variational distributions, with the choice of family playing a critical role in determining the accuracy of the resulting posterior approximation. Simple mean-f…
View article: Joint genotypic and phenotypic outcome modeling improves base editing variant effect quantification
Joint genotypic and phenotypic outcome modeling improves base editing variant effect quantification Open
CRISPR base editing screens are powerful tools for studying disease-associated variants at scale. However, the efficiency and precision of base editing perturbations vary, confounding the assessment of variant-induced phenotypic effects. H…
View article: Joint genotypic and phenotypic outcome modeling improves base editing variant effect quantification
Joint genotypic and phenotypic outcome modeling improves base editing variant effect quantification Open
Data repository for the paper` Joint genotypic and phenotypic outcome modeling improves base editing variant effect measurement`. Includes data required to reproduce analysis in the manuscript through the workflow in https://github.com/pin…
View article: Inferring selection effects in SARS-CoV-2 with Bayesian Viral Allele Selection
Inferring selection effects in SARS-CoV-2 with Bayesian Viral Allele Selection Open
The global effort to sequence millions of SARS-CoV-2 genomes has provided an unprecedented view of viral evolution. Characterizing how selection acts on SARS-CoV-2 is critical to developing effective, long-lasting vaccines and other treatm…
View article: Bayesian Variable Selection in a Million Dimensions
Bayesian Variable Selection in a Million Dimensions Open
Bayesian variable selection is a powerful tool for data analysis, as it offers a principled method for variable selection that accounts for prior information and uncertainty. However, wider adoption of Bayesian variable selection has been …
View article: Analysis of 6.4 million SARS-CoV-2 genomes identifies mutations associated with fitness
Analysis of 6.4 million SARS-CoV-2 genomes identifies mutations associated with fitness Open
Repeated emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants with increased fitness underscores the value of rapid detection and characterization of new lineages. We have developed PyR 0 , a hierarchical Baye…
View article: Inferring selection effects in SARS-CoV-2 with Bayesian Viral Allele Selection
Inferring selection effects in SARS-CoV-2 with Bayesian Viral Allele Selection Open
The global effort to sequence millions of SARS-CoV-2 genomes has provided an unprecedented view of viral evolution. Characterizing how selection acts on SARS-CoV-2 is critical to developing effective, long-lasting vaccines and other treatm…
View article: Surrogate Likelihoods for Variational Annealed Importance Sampling
Surrogate Likelihoods for Variational Annealed Importance Sampling Open
Variational inference is a powerful paradigm for approximate Bayesian inference with a number of appealing properties, including support for model learning and data subsampling. By contrast MCMC methods like Hamiltonian Monte Carlo do not …
View article: Analysis of 6.4 million SARS-CoV-2 genomes identifies mutations associated with fitness
Analysis of 6.4 million SARS-CoV-2 genomes identifies mutations associated with fitness Open
Repeated emergence of SARS-CoV-2 variants with increased fitness necessitates rapid detection and characterization of new lineages. To address this need, we developed PyR 0 , a hierarchical Bayesian multinomial logistic regression model th…
View article: High-Dimensional Bayesian Optimization with Sparse Axis-Aligned Subspaces
High-Dimensional Bayesian Optimization with Sparse Axis-Aligned Subspaces Open
Bayesian optimization (BO) is a powerful paradigm for efficient optimization of black-box objective functions. High-dimensional BO presents a particular challenge, in part because the curse of dimensionality makes it difficult to define---…
View article: Fast Bayesian Variable Selection in Binomial and Negative Binomial Regression
Fast Bayesian Variable Selection in Binomial and Negative Binomial Regression Open
Bayesian variable selection is a powerful tool for data analysis, as it offers a principled method for variable selection that accounts for prior information and uncertainty. However, wider adoption of Bayesian variable selection has been …
View article: Fast Bayesian Variable Selection in Binomial and Negative Binomial\n Regression
Fast Bayesian Variable Selection in Binomial and Negative Binomial\n Regression Open
Bayesian variable selection is a powerful tool for data analysis, as it\noffers a principled method for variable selection that accounts for prior\ninformation and uncertainty. However, wider adoption of Bayesian variable\nselection has be…
View article: Scalable Cross Validation Losses for Gaussian Process Models
Scalable Cross Validation Losses for Gaussian Process Models Open
We introduce a simple and scalable method for training Gaussian process (GP) models that exploits cross-validation and nearest neighbor truncation. To accommodate binary and multi-class classification we leverage Pòlya-Gamma auxiliary vari…
View article: High-Dimensional Bayesian Optimization with Sparse Axis-Aligned Subspaces
High-Dimensional Bayesian Optimization with Sparse Axis-Aligned Subspaces Open
Bayesian optimization (BO) is a powerful paradigm for efficient optimization of black-box objective functions. High-dimensional BO presents a particular challenge, in part because the curse of dimensionality makes it difficult to define --…
View article: Fast Matrix Square Roots with Applications to Gaussian Processes and Bayesian Optimization
Fast Matrix Square Roots with Applications to Gaussian Processes and Bayesian Optimization Open
Matrix square roots and their inverses arise frequently in machine learning, e.g., when sampling from high-dimensional Gaussians $\mathcal{N}(\mathbf 0, \mathbf K)$ or whitening a vector $\mathbf b$ against covariance matrix $\mathbf K$. W…
View article: Deep Sigma Point Processes
Deep Sigma Point Processes Open
We introduce Deep Sigma Point Processes, a class of parametric models inspired by the compositional structure of Deep Gaussian Processes (DGPs). Deep Sigma Point Processes (DSPPs) retain many of the attractive features of (variational) DGP…
View article: Composable Effects for Flexible and Accelerated Probabilistic Programming in NumPyro
Composable Effects for Flexible and Accelerated Probabilistic Programming in NumPyro Open
NumPyro is a lightweight library that provides an alternate NumPy backend to the Pyro probabilistic programming language with the same modeling interface, language primitives and effect handling abstractions. Effect handlers allow Pyro's m…
View article: A Unified Stochastic Gradient Approach to Designing Bayesian-Optimal Experiments
A Unified Stochastic Gradient Approach to Designing Bayesian-Optimal Experiments Open
We introduce a fully stochastic gradient based approach to Bayesian optimal experimental design (BOED). Our approach utilizes variational lower bounds on the expected information gain (EIG) of an experiment that can be simultaneously optim…
View article: Functional Tensors for Probabilistic Programming
Functional Tensors for Probabilistic Programming Open
It is a significant challenge to design probabilistic programming systems that can accommodate a wide variety of inference strategies within a unified framework. Noting that the versatility of modern automatic differentiation frameworks is…
View article: Parametric Gaussian Process Regressors
Parametric Gaussian Process Regressors Open
The combination of inducing point methods with stochastic variational inference has enabled approximate Gaussian Process (GP) inference on large datasets. Unfortunately, the resulting predictive distributions often exhibit substantially un…
View article: Neural Likelihoods for Multi-Output Gaussian Processes
Neural Likelihoods for Multi-Output Gaussian Processes Open
We construct flexible likelihoods for multi-output Gaussian process models that leverage neural networks as components. We make use of sparse variational inference methods to enable scalable approximate inference for the resulting class of…
View article: Variational Estimators for Bayesian Optimal Experimental Design.
Variational Estimators for Bayesian Optimal Experimental Design. Open
View article: Variational Bayesian Optimal Experimental Design
Variational Bayesian Optimal Experimental Design Open
Bayesian optimal experimental design (BOED) is a principled framework for making efficient use of limited experimental resources. Unfortunately, its applicability is hampered by the difficulty of obtaining accurate estimates of the expecte…
View article: Tensor Variable Elimination for Plated Factor Graphs
Tensor Variable Elimination for Plated Factor Graphs Open
A wide class of machine learning algorithms can be reduced to variable elimination on factor graphs. While factor graphs provide a unifying notation for these algorithms, they do not provide a compact way to express repeated structure when…
View article: Closed Form Variational Objectives For Bayesian Neural Networks with a Single Hidden Layer
Closed Form Variational Objectives For Bayesian Neural Networks with a Single Hidden Layer Open
In this note we consider setups in which variational objectives for Bayesian neural networks can be computed in closed form. In particular we focus on single-layer networks in which the activation function is piecewise polynomial (e.g. ReL…
View article: Closed Form Variational Objectives For Bayesian Neural Networks with a\n Single Hidden Layer
Closed Form Variational Objectives For Bayesian Neural Networks with a\n Single Hidden Layer Open
In this note we consider setups in which variational objectives for Bayesian\nneural networks can be computed in closed form. In particular we focus on\nsingle-layer networks in which the activation function is piecewise polynomial\n(e.g. …
View article: Pyro: Deep Universal Probabilistic Programming
Pyro: Deep Universal Probabilistic Programming Open
Pyro is a probabilistic programming language built on Python as a platform for developing advanced probabilistic models in AI research. To scale to large datasets and high-dimensional models, Pyro uses stochastic variational inference algo…
View article: Pathwise Derivatives for Multivariate Distributions
Pathwise Derivatives for Multivariate Distributions Open
We exploit the link between the transport equation and derivatives of expectations to construct efficient pathwise gradient estimators for multivariate distributions. We focus on two main threads. First, we use null solutions of the transp…