TensorFlow Distributions

Abstract

The TensorFlow Distributions library implements a vision of probability theory adapted to the modern deep-learning paradigm of end-to-end differentiable computation. Building on two basic abstractions, it offers flexible building blocks for probabilistic computation. Distributions provide fast, numerically stable methods for generating samples and computing statistics, e.g., log density. Bijectors provide composable volume-tracking transformations with automatic caching. Together these enable modular construction of high dimensional distributions and transformations not possible with previous libraries (e.g., pixelCNNs, autoregressive flows, and reversible residual networks). They are the workhorse behind deep probabilistic programming systems like Edward and empower fast black-box inference in probabilistic models built on deep-network components. TensorFlow Distributions has proven an important part of the TensorFlow toolkit within Google and in the broader deep learning community.

References in corpus (11)

• Auto-Encoding Variational Bayes
• The NumPy array: a structure for efficient numerical computation
• Categorical Reparameterization with Gumbel-Softmax
• Pixel Recurrent Neural Networks
• The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables
• Edward: A library for probabilistic modeling, inference, and criticism
• Learning Stochastic Recurrent Networks
• Gradient Estimation Using Stochastic Computation Graphs
• The Stan Math Library: Reverse-Mode Automatic Differentiation in C++
• ZhuSuan: A Library for Bayesian Deep Learning
• Implicit Causal Models for Genome-wide Association Studies

Cited by in corpus (46)

• Recurrent Neural Networks for Time Series Forecasting: Current Status and Future Directions
• Pyro: Deep Universal Probabilistic Programming
• The Fishyscapes Benchmark: Measuring Blind Spots in Semantic Segmentation
• Neural Network Renormalization Group
• Review: Deep Learning in Electron Microscopy
• Deeply Uncertain: Comparing Methods of Uncertainty Quantification in Deep Learning Algorithms
• Scaling Exact Inference for Discrete Probabilistic Programs
• Probabilistic Deep Learning for Real-Time Large Deformation Simulations
• Non-Intrusive Reduced-Order Modeling Using Uncertainty-Aware Deep Neural Networks and Proper Orthogonal Decomposition: Application to Flood Modeling
• Super-resolution emulator of cosmological simulations using deep physical models
• Non-Gaussian estimates of tensions in cosmological parameters
• Solution of Physics-based Bayesian Inverse Problems with Deep Generative Priors
• INFERNO: Inference-Aware Neural Optimisation
• Explainable Artificial Intelligence for Bayesian Neural Networks: Towards trustworthy predictions of ocean dynamics
• Detection of Gravitational Waves Using Bayesian Neural Networks
• normflows: A PyTorch Package for Normalizing Flows
• Probabilistic Mass Mapping with Neural Score Estimation
• Gaussian process regression + deep neural network autoencoder for probabilistic surrogate modeling in nonlinear mechanics of solids
• Bayesian Hierarchical Probabilistic Forecasting of Intraday Electricity Prices
• Hierarchically modelling Kepler dwarfs and subgiants to improve inference of stellar properties with asteroseismology
• not-MIWAE: Deep Generative Modelling with Missing not at Random Data
• M5 Competition Uncertainty: Overdispersion, distributional forecasting, GAMLSS and beyond
• BayesPerf: Minimizing Performance Monitoring Errors Using Bayesian Statistics
• A Probabilistic Autoencoder for Type Ia Supernovae Spectral Time Series
• Prediction and Uncertainty Quantification of SAFARI-1 Axial Neutron Flux Profiles with Neural Networks
• Bayesian calibration of traffic flow fundamental diagrams using Gaussian processes
• JAGS, NIMBLE, Stan: a detailed comparison among Bayesian MCMC software
• Resampled Priors for Variational Autoencoders
• Neural Density Estimation and Likelihood-free Inference
• Uncertainty-Aware Vehicle Energy Efficiency Prediction using an Ensemble of Neural Networks
• Inferring Signaling Pathways with Probabilistic Programming
• Bayesian Neural Networks at Scale: A Performance Analysis and Pruning Study
• MF-Box: Multi-fidelity and multi-scale emulation for the matter power spectrum
• Gibbs Sampling the Posterior of Neural Networks
• A Priori Uncertainty Quantification of Reacting Turbulence Closure Models using Bayesian Neural Networks
• Supervising the Decoder of Variational Autoencoders to Improve Scientific Utility
• Annealed Flow Transport Monte Carlo
• Uncertainty-Aware Prediction of Battery Energy Consumption for Hybrid Electric Vehicles
• Mixture of Experts Distributional Regression: Implementation Using Robust Estimation with Adaptive First-order Methods
• A Condition Number for Hamiltonian Monte Carlo
• PATHFINDER: Designing Stimuli for Neuromodulation through data-driven inverse estimation of non-linear functions
• JaxSGMC: Modular stochastic gradient MCMC in JAX
• Sparse encoding for more-interpretable feature-selecting representations in probabilistic matrix factorization
• Estimation of redshift and associated uncertainty of Fermi/LAT extra-galactic sources with Deep Learning
• Potential and limitations of machine-learning approaches to inclusive $|V_{ub}|$ determinations
• Classifying different types of solar wind plasma with uncertainty estimations using machine learning