Matthew D. Hoffman
YOU?
Author Swipe
View article: The Impacts of Adaptive Learning Technologies on K-12 Teachers’ Sense of Autonomy, Competence, and Relatedness with Their Students
The Impacts of Adaptive Learning Technologies on K-12 Teachers’ Sense of Autonomy, Competence, and Relatedness with Their Students Open
View article: Running Markov Chain Monte Carlo on Modern Hardware and Software
Running Markov Chain Monte Carlo on Modern Hardware and Software Open
Today, cheap numerical hardware offers huge amounts of parallel computing power, much of which is used for the task of fitting neural networks to data. Adoption of this hardware to accelerate statistical Markov chain Monte Carlo (MCMC) app…
View article: Scalable spatiotemporal prediction with Bayesian neural fields
Scalable spatiotemporal prediction with Bayesian neural fields Open
View article: HaMLET: Human and Machine Learning Effective Teaming
HaMLET: Human and Machine Learning Effective Teaming Open
View article: Distance Learning LDRD: Using Latent Deep Learning Distances to Create Trusted Models
Distance Learning LDRD: Using Latent Deep Learning Distances to Create Trusted Models Open
View article: Nested Rˆ: Assessing the Convergence of Markov Chain Monte Carlo When Running Many Short Chains
Nested Rˆ: Assessing the Convergence of Markov Chain Monte Carlo When Running Many Short Chains Open
Recent developments in parallel Markov chain Monte Carlo (MCMC) algorithms allow us to run thousands of chains almost as quickly as a single chain, using hardware accelerators such as GPUs. While each chain still needs to forget its initia…
View article: Scalable Spatiotemporal Prediction with Bayesian Neural Fields
Scalable Spatiotemporal Prediction with Bayesian Neural Fields Open
Spatiotemporal datasets, which consist of spatially-referenced time series, are ubiquitous in diverse applications, such as air pollution monitoring, disease tracking, and cloud-demand forecasting. As the scale of modern datasets increases…
View article: Reply on RC2
Reply on RC2 Open
Abstract. Some ocean modeling studies have identified a potential tipping point from a low to high basal melt regime beneath the Filchner-Ronne Ice Shelf (FRIS), Antarctica, with significant implications for subsequent Ant…
View article: Reply on RC1
Reply on RC1 Open
Abstract. Some ocean modeling studies have identified a potential tipping point from a low to high basal melt regime beneath the Filchner-Ronne Ice Shelf (FRIS), Antarctica, with significant implications for subsequent Ant…
View article: Robust Inverse Graphics via Probabilistic Inference
Robust Inverse Graphics via Probabilistic Inference Open
How do we infer a 3D scene from a single image in the presence of corruptions like rain, snow or fog? Straightforward domain randomization relies on knowing the family of corruptions ahead of time. Here, we propose a Bayesian approach-dubb…
View article: Training Chain-of-Thought via Latent-Variable Inference
Training Chain-of-Thought via Latent-Variable Inference Open
Large language models (LLMs) solve problems more accurately and interpretably when instructed to work out the answer step by step using a ``chain-of-thought'' (CoT) prompt. One can also improve LLMs' performance on a specific task by super…
View article: Sequential Monte Carlo Learning for Time Series Structure Discovery
Sequential Monte Carlo Learning for Time Series Structure Discovery Open
This paper presents a new approach to automatically discovering accurate models of complex time series data. Working within a Bayesian nonparametric prior over a symbolic space of Gaussian process time series models, we present a novel str…
View article: Impact of Arteriovenous Fistula Compression on Mitral Regurgitation Severity
Impact of Arteriovenous Fistula Compression on Mitral Regurgitation Severity Open
A 57-year-old woman with end-stage renal disease with arteriovenous fistula (AVF) on her left upper extremity was referred for right heart catheterization after diagnosis of mitral regurgitation (MR) by clinical examination and echocardiog…
View article: ProbNeRF: Uncertainty-Aware Inference of 3D Shapes from 2D Images
ProbNeRF: Uncertainty-Aware Inference of 3D Shapes from 2D Images Open
The problem of inferring object shape from a single 2D image is underconstrained. Prior knowledge about what objects are plausible can help, but even given such prior knowledge there may still be uncertainty about the shapes of occluded pa…
View article: Lossy Compression with Gaussian Diffusion
Lossy Compression with Gaussian Diffusion Open
We consider a novel lossy compression approach based on unconditional diffusion generative models, which we call DiffC. Unlike modern compression schemes which rely on transform coding and quantization to restrict the transmitted informati…
View article: Nested $\hat R$: Assessing the convergence of Markov chain Monte Carlo when running many short chains
Nested $\hat R$: Assessing the convergence of Markov chain Monte Carlo when running many short chains Open
Recent developments in parallel Markov chain Monte Carlo (MCMC) algorithms allow us to run thousands of chains almost as quickly as a single chain, using hardware accelerators such as GPUs. While each chain still needs to forget its initia…
View article: Nested $\hat R$: Assessing Convergence for Markov chain Monte Carlo when using many short chains.
Nested $\hat R$: Assessing Convergence for Markov chain Monte Carlo when using many short chains. Open
When using Markov chain Monte Carlo (MCMC) algorithms, we can increase the number of samples either by running longer chains or by running more chains. Practitioners often prefer the first approach because chains need an initial ``warmup''…
View article: What Are Bayesian Neural Network Posteriors Really Like?
What Are Bayesian Neural Network Posteriors Really Like? Open
The posterior over Bayesian neural network (BNN) parameters is extremely high-dimensional and non-convex. For computational reasons, researchers approximate this posterior using inexpensive mini-batch methods such as mean-field variational…
View article: Underspecification Presents Challenges for Credibility in Modern Machine Learning
Underspecification Presents Challenges for Credibility in Modern Machine Learning Open
ML models often exhibit unexpectedly poor behavior when they are deployed in real-world domains. We identify underspecification as a key reason for these failures. An ML pipeline is underspecified when it can return many predictors with eq…
View article: Estimating the Changing Infection Rate of COVID-19 Using Bayesian Models of Mobility
Estimating the Changing Infection Rate of COVID-19 Using Bayesian Models of Mobility Open
In order to prepare for and control the continued spread of the COVID-19 pandemic while minimizing its economic impact, the world needs to be able to estimate and predict COVID-19’s spread. Unfortunately, we cannot directly observe the pre…
View article: RL Unplugged: A Suite of Benchmarks for Offline Reinforcement Learning
RL Unplugged: A Suite of Benchmarks for Offline Reinforcement Learning Open
Offline methods for reinforcement learning have a potential to help bridge the gap between reinforcement learning research and real-world applications. They make it possible to learn policies from offline datasets, thus overcoming concerns…
View article: tfp.mcmc: Modern Markov Chain Monte Carlo Tools Built for Modern Hardware
tfp.mcmc: Modern Markov Chain Monte Carlo Tools Built for Modern Hardware Open
Markov chain Monte Carlo (MCMC) is widely regarded as one of the most important algorithms of the 20th century. Its guarantees of asymptotic convergence, stability, and estimator-variance bounds using only unnormalized probability function…
View article: Hamiltonian Monte Carlo Swindles
Hamiltonian Monte Carlo Swindles Open
Hamiltonian Monte Carlo (HMC) is a powerful Markov chain Monte Carlo (MCMC) algorithm for estimating expectations with respect to continuous un-normalized probability distributions. MCMC estimators typically have higher variance than class…
View article: Automatically Batching Control-Intensive Programs for Modern Accelerators
Automatically Batching Control-Intensive Programs for Modern Accelerators Open
We present a general approach to batching arbitrary computations for accelerators such as GPUs. We show orders-of-magnitude speedups using our method on the No U-Turn Sampler (NUTS), a workhorse algorithm in Bayesian statistics. The centra…
View article: Automatic Reparameterisation of Probabilistic Programs
Automatic Reparameterisation of Probabilistic Programs Open
Probabilistic programming has emerged as a powerful paradigm in statistics, applied science, and machine learning: by decoupling modelling from inference, it promises to allow modellers to directly reason about the processes generating dat…
View article: NeuTra-lizing Bad Geometry in Hamiltonian Monte Carlo Using Neural Transport
NeuTra-lizing Bad Geometry in Hamiltonian Monte Carlo Using Neural Transport Open
Hamiltonian Monte Carlo is a powerful algorithm for sampling from difficult-to-normalize posterior distributions. However, when the geometry of the posterior is unfavorable, it may take many expensive evaluations of the target distribution…
View article: Autoconj: Recognizing and Exploiting Conjugacy Without a Domain-Specific Language
Autoconj: Recognizing and Exploiting Conjugacy Without a Domain-Specific Language Open
Deriving conditional and marginal distributions using conjugacy relationships can be time consuming and error prone. In this paper, we propose a strategy for automating such derivations. Unlike previous systems which focus on relationships…
View article: Autoconj: Recognizing and Exploiting Conjugacy Without a Domain-Specific\n Language
Autoconj: Recognizing and Exploiting Conjugacy Without a Domain-Specific\n Language Open
Deriving conditional and marginal distributions using conjugacy relationships\ncan be time consuming and error prone. In this paper, we propose a strategy for\nautomating such derivations. Unlike previous systems which focus on\nrelationsh…
View article: Simple, Distributed, and Accelerated Probabilistic Programming
Simple, Distributed, and Accelerated Probabilistic Programming Open
We describe a simple, low-level approach for embedding probabilistic programming in a deep learning ecosystem. In particular, we distill probabilistic programming down to a single abstraction---the random variable. Our lightweight implemen…
View article: Characterizing User Skills from Application Usage Traces with Hierarchical Attention Recurrent Networks
Characterizing User Skills from Application Usage Traces with Hierarchical Attention Recurrent Networks Open
Predicting users’ proficiencies is a critical component of AI-powered personal assistants. This article introduces a novel approach for the prediction based on users’ diverse, noisy, and passively generated application usage histories. We …