Improving black-box optimization in VAE latent space using decoder uncertainty

Abstract

Optimization in the latent space of variational autoencoders is a promising approach to generate high-dimensional discrete objects that maximize an expensive black-box property (e.g., drug-likeness in molecular generation, function approximation with arithmetic expressions). However, existing methods lack robustness as they may decide to explore areas of the latent space for which no data was available during training and where the decoder can be unreliable, leading to the generation of unrealistic or invalid objects. We propose to leverage the epistemic uncertainty of the decoder to guide the optimization process. This is not trivial though, as a naive estimation of uncertainty in the high-dimensional and structured settings we consider would result in high estimator variance. To solve this problem, we introduce an importance sampling-based estimator that provides more robust estimates of epistemic uncertainty. Our uncertainty-guided optimization approach does not require modifications of the model architecture nor the training process. It produces samples with a better trade-off between black-box objective and validity of the generated samples, sometimes improving both simultaneously. We illustrate these advantages across several experimental settings in digit generation, arithmetic expression approximation and molecule generation for drug design.

References in corpus (9)

• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Auto-Encoding Variational Bayes
• Stochastic Backpropagation and Approximate Inference in Deep Generative Models
• Convolutional Networks on Graphs for Learning Molecular Fingerprints
• Junction Tree Variational Autoencoder for Molecular Graph Generation
• Learning Deep Generative Models of Graphs
• Detecting Adversarial Samples from Artifacts
• Syntax-Directed Variational Autoencoder for Structured Data
• Understanding Measures of Uncertainty for Adversarial Example Detection

Cited by in corpus (5)

• High-Dimensional Bayesian Optimisation with Variational Autoencoders and Deep Metric Learning
• Combining Latent Space and Structured Kernels for Bayesian Optimization over Combinatorial Spaces
• HyAR: Addressing Discrete-Continuous Action Reinforcement Learning via Hybrid Action Representation
• Leveraging Active Subspaces to Capture Epistemic Model Uncertainty in Deep Generative Models for Molecular Design
• High-Dimensional Bayesian Optimization via Semi-Supervised Learning with Optimized Unlabeled Data Sampling