DEBOSH: Deep Bayesian Shape Optimization
Nikita Durasov, Artem Lukoyanov, Jonathan Donier, Pascal Fua
TL;DR
This paper introduces DEBOSH, a novel Bayesian shape optimization method using deep neural networks that effectively estimates uncertainty, enabling better exploration of shape space and improved design quality.
Contribution
It presents a new uncertainty estimation approach for neural networks in shape optimization, enhancing Bayesian Optimization performance.
Findings
Outperforms state-of-the-art shape optimization methods
Enables reliable uncertainty estimation for neural networks
Achieves higher quality shapes in optimization tasks
Abstract
Graph Neural Networks (GNNs) can predict the performance of an industrial design quickly and accurately and be used to optimize its shape effectively. However, to fully explore the shape space, one must often consider shapes deviating significantly from the training set. For these, GNN predictions become unreliable, something that is often ignored. For optimization techniques relying on Gaussian Processes, Bayesian Optimization (BO) addresses this issue by exploiting their ability to assess their own accuracy. Unfortunately, this is harder to do when using neural networks because standard approaches to estimating their uncertainty can entail high computational loads and reduced model accuracy. Hence, we propose a novel uncertainty-based method tailored to shape optimization. It enables effective BO and increases the quality of the resulting shapes beyond that of state-of-the-art…
Peer Reviews
Decision·ICLR 2024 Conference Withdrawn Submission
- The interdisciplinary approach to shape optimization at the intersection of engineering design and machine learning is very appealing. - The idea of physically-driven uncertainty prediction in the context of shape simulation is inspiring. - The paper is well-written and well-organized.
- The framework is limited in the sense that it bounds itself to a set of pre-existing shapes in the dataset. In fact, the main question that is addressed is how one can find the most promising shapes to be passed to physical simulator. This is a less strong achievement than a full-fledged BO which could 'synthesis' new shapes. - The above point begs the question of how the cost of training and optimization are amortized during later inference stages. At least in the case of 1500 airfoils, I f
The paper tackles an important topic, namely uncertainty quantification of surrogate models for geometric design optimization. They have conducted a set of experiments on a dataset of airfoil shapes, and car shapes. The car dataset appears to be particularly challenging since the latent geometric parameter is high dimensional. The model with the entrant GNN obtains good experimental results in predicting the drag and lift-to-drag coefficients, especially when there are a few training samples.
First, the paper is really dense and quite hard to read. I would suggest some rewriting to make it less compact. Then, I am not convinced by the approach taken by the authors w.r.t the output of the surrogate model. Here, the surrogate model learns to predict both the surface pressure and the metric value. Even though this metric is computed from approximate pressure values with another neural network, it still breaks the physical integration between the pressure field and the different forces.
Results are good. Moderately good description of the process. Good for CFD community.
Only one equation in manuscript - indicates lack of soundness and novelty. Do not agree that use of Reentrant GNN with Bayesian optimization is sufficient for ICLR level. Where are failure cases illustrated and discussed ?
This paper attempts to propose a new pipeline to address the challenges encountered in practical industrial manufacturing, and this research direction is meaningful. Additionally, the authors present a comprehensive technical approach and conduct experiments on relevant data. Overall, the paper is well-written and easy to follow.
The proposed approach in this paper seems reasonable, but I still have some concerns. The main issue is that the current design changes are entirely optimized by the network adaptively. While this may indeed yield a higher metric on the designed acquisition function, I am not entirely certain if this metric truly reflects real-world conditions. On one hand, the optimization direction may not necessarily align with the expectations of industrial design (the metric itself may only capture one as
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
Topics3D Shape Modeling and Analysis · Computer Graphics and Visualization Techniques · Visual Attention and Saliency Detection