An official website of the United States government
Bayesian optimization (BO) is a powerful paradigm for optimizing expensive black-box functions. Traditional BO methods typically rely on separate hand-crafted acquisition functions and surrogate models for the underlying function, and often operate in a myopic manner. In this paper, we propose a novel direct regret optimization approach that jointly learns the optimal model and non-myopic acquisition by distilling from a set of candidate models and acquisitions, and explicitly targets minimizing the multi-step regret. Our framework leverages an ensemble of Gaussian Processes (GPs) with varying hyperparameters to generate simulated BO trajectories, each guided by an acquisition function chosen from a pool of conventional choices, until a Bayesian early stop criterion is met. These simulated trajectories, capturing multi-step exploration strategies, are used to train an end-to-end decision transformer that directly learns to select next query points aimed at improving the ultimate objective. We further adopt a dense training–sparse learning paradigm: The decision transformer is trained offline with abundant simulated data sampled from ensemble GPs and acquisitions, while a limited number of real evaluations refine the GPs online. Experimental results on synthetic and real-world benchmarks suggest that our method consistently outperforms BO baselines, achieving lower simple regret and demonstrating more robust exploration in high-dimensional or noisy settings.
more »
« less
Thermal-Aware SoC Macro Placement and Multi-chip Module Design Optimization with Bayesian Optimization
More Like this
-
-
The rapid advancement of 5G technology necessitates the development of efficient thermal management solutions to handle the increased heat dissipation demands of high-power electronic components. This study presents an optimization strategy for a microchannel cold plate designed for a prototype 5G front-end system, featuring four 22-Watt chips as heat sources. The cold plate, constructed from aluminum, incorporates multiple rectangular flow channels evenly spaced to facilitate uniform heat distribution, with an inlet runner. The primary objective of this study is to optimize the geometry of the flow channels and the coolant mass flow rate at the runner entrance to minimize entropy generation, thereby enhancing the heat dissipation capability of the cold plate while minimizing pressure drop. Given these challenges, this study aims to develop an optimization strategy for cold plate design. This research applies Bayesian Optimization (BO), and Response Surface Methodology (RSM) paired with Genetic Algorithm (GA), and FMINCON (sequential quadratic programming, a built-in optimizer of MATLAB). These methods are utilized to fine-tune the channel dimensions and coolant flow rate, and the data that is used to evaluate entropy of the system is obtained from conjugate heat transfer simulations solved by Ansys Fluent. By using the Gaussian Process model to build response surface and predicting function of entropy generation, the results indicate that BO outperforms RSM paired with GA and FMINCON in terms of entropy reduction with same number of samples.more » « less
