eCNNTO: A Highly Generalizable ConvNet for Accelerating Topology Optimization

A new convolutional neural network architecture called eCNNTO can accelerate density-based topology optimization by up to 97 percent in three-dimensional problems, according to a preprint posted to arXiv on June 18, 2026 [1][2]. The method, detailed in a paper submitted to the artificial intelligence section of the open-access repository, addresses a long-standing bottleneck in topology optimization: the large number of iterations required when finite element analysis is performed on dense meshes for high-resolution designs [1][2]. The authors build on prior work by Kallioras et al. (2020), which used a Deep Belief Network to predict near-optimal element density from early-stage history and skip most iterations [1][2]. That earlier approach lacked spatial correlations among neighboring elements and could produce disconnected features in the final structure [1][2]. eCNNTO replaces the Deep Belief Network with a convolutional neural network that incorporates residual connections, preserving spatial relationships across elements [1][2]. The researchers also introduced a training strategy that uses density histories from the final stage of optimization rather than early iterations, a change the paper states helps reduce the required training data size [1][2]. The model can be trained on a small dataset and generalizes to problems with different boundary conditions, loading cases, design-domain geometries, mesh resolutions, and non-design domains [1][2]. In two-dimensional examples, eCNNTO achieved up to a 90 percent reduction in iterations; in three dimensions, the reduction reached 97 percent [1][2]. arXiv, which hosts the preprint, was founded in 1991 and passed the two-million-article milestone by the end of 2021 [6]. As of November 2024, the repository was receiving about 24,000 submissions per month across fields including computer science, physics, and mathematics [6]. Papers on arXiv are moderated but not peer-reviewed before posting [6]. The eCNNTO manuscript appeared alongside the platform's arXivLabs framework, which allows third-party collaborators to build experimental tools on article pages under guidelines that require adherence to openness and user-data privacy [4][5]. The framework, formalized in 2020, includes tools such as the Bibliographic Explorer and CORE Recommender that help readers navigate citation trees and discover related open-access research [4][5].