Improved Off-policy Reinforcement Learning in Biological Sequence Design

Designing biological sequences with desired properties is a significant challenge due to the combinatorially vast search space and the high cost of evaluating each candidate sequence. To address these challenges, reinforcement learning (RL) methods, such as GFlowNets, utilize proxy models for rapid reward evaluation and annotated data for policy training. Although these approaches have shown promise in generating diverse and novel sequences, the limited training data relative to the vast search space often leads to the misspecification of proxy for out-of-distribution inputs. We introduce $\delta$-Conservative Search, a novel off-policy search method for training GFlowNets designed to improve robustness against proxy misspecification. The key idea is to incorporate conservativeness, controlled by parameter $\delta$, to constrain the search to reliable regions. Specifically, we inject noise into high-score offline sequences by randomly masking tokens with a Bernoulli distribution of parameter $\delta$ and then denoise masked tokens using the GFlowNet policy. Additionally, $\delta$ is adaptively adjusted based on the uncertainty of the proxy model for each data point. This enables the reflection of proxy uncertainty to determine the level of conservativeness. Experimental results demonstrate that our method consistently outperforms existing machine learning methods in discovering high-score sequences across diverse tasks-including DNA, RNA, protein, and peptide design-especially in large-scale scenarios.

Adaptive teachers for amortized samplers

Amortized inference is the task of training a parametric model, such as a neural network, to approximate a distribution with a given unnormalized density where exact sampling is intractable. When sampling is implemented as a sequential decision-making process, reinforcement learning (RL) methods, such as generative flow networks, can be used to train the sampling policy. Off-policy RL training facilitates the discovery of diverse, high-reward candidates, but existing methods still face challenges in efficient exploration. We propose to use an adaptive training distribution (the Teacher) to guide the training of the primary amortized sampler (the Student) by prioritizing high-loss regions. The Teacher, an auxiliary behavior model, is trained to sample high-error regions of the Student and can generalize across unexplored modes, thereby enhancing mode coverage by providing an efficient training curriculum. We validate the effectiveness of this approach in a synthetic environment designed to present an exploration challenge, two diffusion-based sampling tasks, and four biochemical discovery tasks demonstrating its ability to improve sample efficiency and mode coverage.

An Offline Meta Black-box Optimization Framework for Adaptive Design of Urban Traffic Light Management Systems

Complex urban road networks with high vehicle occupancy frequently face severe traffic congestion. Designing an effective strategy for managing multiple traffic lights plays a crucial role in managing congestion. However, most current traffic light management systems rely on human-crafted decisions, which may not adapt well to diverse traffic patterns. In this paper, we delve into two pivotal design components of the traffic light management system that can be dynamically adjusted to various traffic conditions: phase combination and phase time allocation. While numerous studies have sought an efficient strategy for managing traffic lights, most of these approaches consider a fixed traffic pattern and are limited to relatively small road networks. To overcome these limitations, we introduce a novel and practical framework to formulate the optimization of such design components using an offline meta black-box optimization. We then present a simple yet effective method to efficiently find a solution for the aforementioned problem. In our framework, we first collect an offline meta dataset consisting of pairs of design choices and corresponding congestion measures from various traffic patterns. After collecting the dataset, we employ the Attentive Neural Process (ANP) to predict the impact of the proposed design on congestion across various traffic patterns with well-calibrated uncertainty. Finally, Bayesian optimization, with ANP as a surrogate model, is utilized to find an optimal design for unseen traffic patterns through limited online simulations. Our experiment results show that our method outperforms state-of-the-art baselines on complex road networks in terms of the number of waiting vehicles. Surprisingly, the deployment of our method into a real-world traffic system was able to improve traffic throughput by 4.80\% compared to the original strategy.

Guided Trajectory Generation with Diffusion Models for Offline Model-based Optimization

Optimizing complex and high-dimensional black-box functions is ubiquitous in science and engineering fields. Unfortunately, the online evaluation of these functions is restricted due to time and safety constraints in most cases. In offline model-based optimization (MBO), we aim to find a design that maximizes the target function using only a pre-existing offline dataset. While prior methods consider forward or inverse approaches to address the problem, these approaches are limited by conservatism and the difficulty of learning highly multi-modal mappings. Recently, there has been an emerging paradigm of learning to improve solutions with synthetic trajectories constructed from the offline dataset. In this paper, we introduce a novel conditional generative modeling approach to produce trajectories toward high-scoring regions. First, we construct synthetic trajectories toward high-scoring regions using the dataset while injecting locality bias for consistent improvement directions. Then, we train a conditional diffusion model to generate trajectories conditioned on their scores. Lastly, we sample multiple trajectories from the trained model with guidance to explore high-scoring regions beyond the dataset and select high-fidelity designs among generated trajectories with the proxy function. Extensive experiment results demonstrate that our method outperforms competitive baselines on Design-Bench and its practical variants. The code is publicly available in \texttt{https://github.com/dbsxodud-11/GTG}.

GTA: Generative Trajectory Augmentation with Guidance for Offline Reinforcement Learning

Offline Reinforcement Learning (Offline RL) presents challenges of learning effective decision-making policies from static datasets without any online interactions. Data augmentation techniques, such as noise injection and data synthesizing, aim to improve Q-function approximation by smoothing the learned state-action region. However, these methods often fall short of directly improving the quality of offline datasets, leading to suboptimal results. In response, we introduce \textbf{GTA}, Generative Trajectory Augmentation, a novel generative data augmentation approach designed to enrich offline data by augmenting trajectories to be both high-rewarding and dynamically plausible. GTA applies a diffusion model within the data augmentation framework. GTA partially noises original trajectories and then denoises them with classifier-free guidance via conditioning on amplified return value. Our results show that GTA, as a general data augmentation strategy, enhances the performance of widely used offline RL algorithms in both dense and sparse reward settings. Furthermore, we conduct a quality analysis of data augmented by GTA and demonstrate that GTA improves the quality of the data. Our code is available at https://github.com/Jaewoopudding/GTA

Learning to Scale Logits for Temperature-Conditional GFlowNets

GFlowNets are probabilistic models that learn a stochastic policy that sequentially generates compositional structures, such as molecular graphs. They are trained with the objective of sampling such objects with probability proportional to the object's reward. Among GFlowNets, the temperature-conditional GFlowNets represent a family of policies indexed by temperature, and each is associated with the correspondingly tempered reward function. The major benefit of temperature-conditional GFlowNets is the controllability of GFlowNets' exploration and exploitation through adjusting temperature. We propose Learning to Scale Logits for temperature-conditional GFlowNets (LSL-GFN), a novel architectural design that greatly accelerates the training of temperature-conditional GFlowNets. It is based on the idea that previously proposed temperature-conditioning approaches introduced numerical challenges in the training of the deep network because different temperatures may give rise to very different gradient profiles and ideal scales of the policy's logits. We find that the challenge is greatly reduced if a learned function of the temperature is used to scale the policy's logits directly. We empirically show that our strategy dramatically improves the performances of GFlowNets, outperforming other baselines, including reinforcement learning and sampling methods, in terms of discovering diverse modes in multiple biochemical tasks.

Local Search GFlowNets

Generative Flow Networks (GFlowNets) are amortized sampling methods that learn a distribution over discrete objects proportional to their rewards. GFlowNets exhibit a remarkable ability to generate diverse samples, yet occasionally struggle to consistently produce samples with high rewards due to over-exploration on wide sample space. This paper proposes to train GFlowNets with local search which focuses on exploiting high rewarded sample space to resolve this issue. Our main idea is to explore the local neighborhood via destruction and reconstruction guided by backward and forward policies, respectively. This allows biasing the samples toward high-reward solutions, which is not possible for a typical GFlowNet solution generation scheme which uses the forward policy to generate the solution from scratch. Extensive experiments demonstrate a remarkable performance improvement in several biochemical tasks. Source code is available: \url{https://github.com/dbsxodud-11/ls_gfn}.