Differentiable Expected Hypervolume Improvement for Parallel Multi-Objective Bayesian Optimization

Abstract

In many real-world scenarios, decision makers seek to efficiently optimize multiple competing objectives in a sample-efficient fashion. Multi-objective Bayesian optimization (BO) is a common approach, but many of the best-performing acquisition functions do not have known analytic gradients and suffer from high computational overhead. We leverage recent advances in programming models and hardware acceleration for multi-objective BO using Expected Hypervolume Improvement (EHVI)---an algorithm notorious for its high computational complexity. We derive a novel formulation of q-Expected Hypervolume Improvement (qEHVI), an acquisition function that extends EHVI to the parallel, constrained evaluation setting. qEHVI is an exact computation of the joint EHVI of q new candidate points (up to Monte-Carlo (MC) integration error). Whereas previous EHVI formulations rely on gradient-free acquisition optimization or approximated gradients, we compute exact gradients of the MC estimator via auto-differentiation, thereby enabling efficient and effective optimization using first-order and quasi-second-order methods. Our empirical evaluation demonstrates that qEHVI is computationally tractable in many practical scenarios and outperforms state-of-the-art multi-objective BO algorithms at a fraction of their wall time.

arXiv commentTo appear in Advances in Neural Information Processing Systems 33, 2020. Code is available at https://github.com/pytorch/botorch

References in corpus (11)

• An Easy-to-use Real-world Multi-objective Optimization Problem Suite
• Bayesian Optimization with Unknown Constraints
• BoTorch: A Framework for Efficient Monte-Carlo Bayesian Optimization
• A Bayesian approach to constrained single- and multi-objective optimization
• The Parallel Knowledge Gradient Method for Batch Bayesian Optimization
• Predictive Entropy Search for Multi-objective Bayesian Optimization with Constraints
• Maximizing acquisition functions for Bayesian optimization
• Parallel Bayesian Global Optimization of Expensive Functions
• The reparameterization trick for acquisition functions
• Targeting Solutions in Bayesian Multi-Objective Optimization: Sequential and Batch Versions
• Parallel Predictive Entropy Search for Multi-objective Bayesian Optimization with Constraints

Cited by in corpus (23)

• Multi-Objective Bayesian Optimization for Accelerator Tuning
• Bag of Baselines for Multi-objective Joint Neural Architecture Search and Hyperparameter Optimization
• Multi-Objective Bayesian Active Learning for MeV-ultrafast electron diffraction
• A Data Driven Sequential Learning Framework to Accelerate and Optimize Multi-Objective Manufacturing Decisions
• Multi-Objective Bayesian Optimization over High-Dimensional Search Spaces
• Leveraging Trust for Joint Multi-Objective and Multi-Fidelity Optimization
• Pareto Front-Diverse Batch Multi-Objective Bayesian Optimization
• DARTH: Declarative Recall Through Early Termination for Approximate Nearest Neighbor Search
• Constrained composite Bayesian optimization for rational synthesis of polymeric particles
• Ranking over Regression for Bayesian Optimization and Molecule Selection
• Combining Latent Space and Structured Kernels for Bayesian Optimization over Combinatorial Spaces
• Perceptual-Distortion Balanced Image Super-Resolution is a Multi-Objective Optimization Problem
• Mono-surrogate vs Multi-surrogate in Multi-objective Bayesian Optimisation
• Bayesian Optimization Algorithms for Accelerator Physics
• Bayesian optimisation of poloidal field coil positions in tokamaks
• Bandit Pareto Set Identification: the Fixed Budget Setting
• Multi-objective Optimization by Learning Space Partitions
• Multi-Objective Learning to Predict Pareto Fronts Using Hypervolume Maximization
• Multi-objective Bayesian optimization of ferroelectric materials with interfacial control for memory and energy storage applications
• Advancing the Pareto front for thin-film materials using a self-driving laboratory
• Choice functions based multi-objective Bayesian optimisation
• PASTO: Strategic Parameter Optimization in Recommendation Systems -- Probabilistic is Better than Deterministic
• Achieving Diversity in Objective Space for Sample-efficient Search of Multiobjective Optimization Problems