mlsanityData to 7 Oct 2026

Deep Adaptive Design: Amortizing Sequential Bayesian Experimental Design

arXiv 2103.02438 ↗First posted 3 Mar 2021stat.ML cs.AI cs.LGcited by 103

Not on the map. Why?

Abstract (arXiv)

We introduce Deep Adaptive Design (DAD), a method for amortizing the cost of adaptive Bayesian experimental design that allows experiments to be run in real-time. Traditional sequential Bayesian optimal experimental design approaches require substantial computation at each stage of the experiment. This makes them unsuitable for most real-world applications, where decisions must typically be made quickly. DAD addresses this restriction by learning an amortized design network upfront and then using this to rapidly run (multiple) adaptive experiments at deployment time. This network represents a design policy which takes as input the data from previous steps, and outputs the next design using a single forward pass; these design decisions can be made in milliseconds during the live experiment. To train the network, we introduce contrastive information bounds that are suitable objectives for the sequential setting, and propose a customized network architecture that exploits key symmetries. We demonstrate that DAD successfully amortizes the process of experimental design, outperforming alternative strategies on a number of problems.

Key facts. Cited by 103 papers in the map (data through 7 Oct 2026). Builds on 25 of its 66 references that are in the map. First posted on arXiv 3 Mar 2021.

Builds on

25 of 66 references.

References · 25

  1. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  2. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  3. 2018Representation Learning with Contrastive Predictive Codingcited by 7,662
  4. 2013Auto-Encoding Variational Bayescited by 3,980
  5. 2017Deep Setscited by 1,394
  6. 2015Automatic differentiation in machine learning: a surveycited by 846
  7. 2019On Variational Bounds of Mutual Informationcited by 550
  8. 2018Conditional Neural Processescited by 421
  9. 2018Pyro: Deep Universal Probabilistic Programmingcited by 379
  10. 2020Planning to Explore via Self-Supervised World Modelscited by 365
  11. 2018Neural Processescited by 307
  12. 2019Monte Carlo Gradient Estimation in Machine Learningcited by 281
13 more
  1. 2018Inference Suboptimality in Variational Autoencoderscited by 131
  2. 2015Gradient Estimation Using Stochastic Computation Graphscited by 120
  3. 2016Towards a Neural Statisticiancited by 119
  4. 2019Variational Bayesian Optimal Experimental Designcited by 116
  5. 2017On Nesting Monte Carlo Estimatorscited by 82
  6. 2017REBAR: Low-variance, unbiased gradient estimates for discrete latent variable modelscited by 78
  7. 2019A Unified Stochastic Gradient Approach to Designing Bayesian-Optimal Experimentscited by 58
  8. 2020Bayesian Experimental Design for Implicit Models by Mutual Information Neural Estimationcited by 56
  9. 2015GLASSES: Relieving The Myopia Of Bayesian Optimisationcited by 38
  10. 2019BINOCULARS for Efficient, Nonmyopic Sequential Experimental Designcited by 35
  11. 2016Sequential Bayesian optimal experimental design via approximate dynamic programmingcited by 27
  12. 2020Sequential Bayesian Experimental Design for Implicit Models via Mutual Informationcited by 23
  13. 2020Unbiased MLMC stochastic gradient-based optimization of Bayesian experimental designscited by 19

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 6; 2022: 11; 2023: 13; 2024: 14; 2025: 25; 2026: 34.

103 papers.

Sort:

By area:

  1. 2023Modern Bayesian Experimental Designcited by 121
  2. 2021Active Testing: Sample-Efficient Model Evaluationcited by 82
  3. 2023GFlowNets for AI-Driven Scientific Discoverycited by 63
  4. 2021Implicit Deep Adaptive Design: Policy-Based Experimental Design without Likelihoodscited by 61
  5. 2022Interventions, Where and How? Experimental Design for Causal Models at Scalecited by 49
  6. 2022Optimizing Sequential Experimental Design with Deep Reinforcement Learningcited by 45
  7. 2024The Essential Role of Causality in Foundation World Models for Embodied AIcited by 32
  8. 2021Gradient-based Bayesian Experimental Design for Implicit Models using Mutual Information Lower Boundscited by 26
  9. 2022Active Surrogate Estimators: An Active Learning Approach to Label-Efficient Model Evaluationcited by 26
  10. 2025BED-LLM: Intelligent Information Gathering with LLMs and Bayesian Experimental Designcited by 21
  11. 2024Amortized Bayesian Experimental Design for Decision-Makingcited by 20
  12. 2022Active Exploration via Experiment Design in Markov Chainscited by 19
38 more
  1. 2021Tight Mutual Information Estimation With Contrastive Fenchel-Legendre Optimizationcited by 18
  2. 2022Combining Multi-Fidelity Modelling and Asynchronous Batch Bayesian Optimizationcited by 16
  3. 2022Policy-Based Bayesian Experimental Design for Non-Differentiable Implicit Modelscited by 15
  4. 2023Differentiable Multi-Target Causal Bayesian Experimental Designcited by 15
  5. 2025BoxingGym: Benchmarking Progress in Automated Experimental Design and Model Discoverycited by 15
  6. 2021Active Learning in Bayesian Neural Networks with Balanced Entropy Learning Principlecited by 14
  7. 2023Leveraging Self-Consistency for Data-Efficient Amortized Bayesian Inferencecited by 14
  8. 2023Black-Box Batch Active Learning for Regressioncited by 12
  9. 2024Nesting Particle Filters for Experimental Design in Dynamical Systemscited by 12
  10. 2025Step-DAD: Semi-Amortized Policy-Based Bayesian Experimental Designcited by 12
  11. 2024Amortized Active Causal Induction with Deep Reinforcement Learningcited by 10
  12. 2024Bayesian Experimental Design via Contrastive Diffusionscited by 9
  13. 2023Variational Sequential Optimal Experimental Design using Reinforcement Learningcited by 8
  14. 2025ALINE: Joint Amortization for Bayesian Inference and Active Data Acquisitioncited by 8
  15. 2021Deep Bayesian Active Learning for Accelerating Stochastic Simulationcited by 7
  16. 2022Design Amortization for Bayesian Optimal Experimental Designcited by 7
  17. 2024Advancing Deep Active Learning & Data Subset Selection: Unifying Principles with Information-Theory Intuitionscited by 7
  18. 2025PABBO: Preferential Amortized Black-Box Optimizationcited by 7
  19. 2024Unraveling the Interplay between Carryover Effects and Reward Autocorrelations in Switchback Experimentscited by 6
  20. 2026Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world modelscited by 6
  21. 2023Statistically Efficient Bayesian Sequential Experiment Design via Reinforcement Learning with Cross-Entropy Estimatorscited by 5
  22. 2024PASOA- PArticle baSed Bayesian Optimal Adaptive designcited by 5
  23. 2025Active Learning of Model Discrepancy with Bayesian Experimental Designcited by 5
  24. 2025Balancing Interference and Correlation in Spatial Experimental Designs: A Causal Graph Cut Approachcited by 5
  25. 2025In-Context Multi-Objective Optimizationcited by 5
  26. 2022Bayesian Active Learning for Discrete Latent Variable Modelscited by 4
  27. 2023CO-BED: Information-Theoretic Contextual Optimization via Bayesian Experimental Designcited by 4
  28. 2023Online simulator-based experimental design for cognitive model selectioncited by 3
  29. 2023Experimentation Platforms Meet Reinforcement Learning: Bayesian Sequential Decision-Making for Continuous Monitoringcited by 3
  30. 2024Recursive Nested Filtering for Efficient Amortized Bayesian Experimental Designcited by 3
  31. 2025Amortized Safe Active Learning for Real-Time Data Acquisition: Pretrained Neural Policies From Simulated Nonparametric Functionscited by 3
  32. 2025Performance Comparisons of Reinforcement Learning Algorithms for Sequential Experimental Designcited by 3
  33. 2026ATLAS: Active Theory Learning for Automated Sciencecited by 3
  34. 2023Adaptive Experimentation at Scale: A Computational Framework for Flexible Batchescited by 2
  35. 2024Amortized nonmyopic active search via deep imitation learningcited by 2
  36. 2025PIED: Physics-Informed Experimental Design for Inverse Problemscited by 2
  37. 2025On Efficient Bayesian Exploration in Model-Based Reinforcement Learningcited by 2
  38. 2025FraPPE: Fast and Efficient Preference-based Pure Explorationcited by 2

50 most cited of 103 All 103 (JSON)