Bayesian Optimization of Function Networks
Dec 6, 2021
Speakers
Raúl Astudillo
Speaker · 0 followers
Peter I. Frazier
Speaker · 0 followers
About
We consider Bayesian optimization of the output of a network of functions, where each function takes as input the output of its parent nodes, and where the network takes significant time to evaluate. Such problems arise, for example, in reinforcement learning, engineering design, and manufacturing. While the standard Bayesian optimization approach observes only the final output, our approach delivers greater query efficiency by leveraging information that the former ignores: intermediate output within the network. This is achieved by modeling the nodes of the network using Gaussian processes and choosing the points to evaluate using, as our acquisition function, the expected improvement computed with respect to the implied posterior on the objective function. Although the non-Gaussian nature of this posterior prevents computing our acquisition function in closed form, we maximize it using a sample average approximation approach. In addition, we show that our method is asymptotically consistent, meaning that it finds a globally optimal solution as the number of evaluations grows to infinity, thus generalizing previously known convergence results for the expected improvement acquisition function. This holds even though it might not measure densely, instead leveraging problem structure to leave regions unexplored. Finally, our approach dramatically outperforms standard BO methods in several synthetic and real-world problems.We consider Bayesian optimization of the output of a network of functions, where each function takes as input the output of its parent nodes, and where the network takes significant time to evaluate. Such problems arise, for example, in reinforcement learning, engineering design, and manufacturing. While the standard Bayesian optimization approach observes only the final output, our approach delivers greater query efficiency by leveraging information that the former ignores: intermediate output…
Organizer
NeurIPS 2021
Account · 1.9k followers
About NeurIPS 2021
Neural Information Processing Systems (NeurIPS) is a multi-track machine learning and computational neuroscience conference that includes invited talks, demonstrations, symposia and oral and poster presentations of refereed papers. Following the conference, there are workshops which provide a less formal setting.
Like the format? Trust SlidesLive to capture your next event!
Professional recording and live streaming, delivered globally.
Sharing
Recommended Videos
Presentations on similar topic, category or speaker
Addressing Algorithmic Disparity and Performance Inconsistency in Federated Learning
Watch later
Sen Cui, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Generative Occupancy Fields for 3D Surface-Aware Image Synthesis
Watch later
Xudong Xu, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Graph Robustness Benchmark: Benchmarking the Adversarial Robustness of Graph Machine Learning
Watch later
Qinkai Zheng, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Learning and Decision-Making with Strategic Feedback
Watch later
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
A Comprehensively Tight Analysis of Gradient Descent for PCA
Watch later
Zhiqiang Xu, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Self-Supervised Learning: Self-Prediction and Contrastive Learning
Watch later
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%