arXiv Machine Learning
Jul 7
Generative Inverse Design with Abstention via Diagonal Flow Matching
arXiv:2603. 15925v2 Announce Type: replace Abstract: Inverse design aims to find design parameters $x$ achieving target performance $y^*$.
Stream-Based Active Learning with Cooperative Neural Networks for Data-Efficient Partial Inverse Design: An Automotive Glass Run Channel Case Study
The paper introduces CoNN-AL, a framework that combines stream-based active learning with a Cooperative Neural Network and Denoising Autoencoder to efficiently perform partial inverse design in engineering. By estimating predictive uncertainty with Monte Carlo dropout, the method selectively labels the most informative samples in real time, dramatically reducing the number of required simulations. Applied to an automotive glass run channel dataset of over 900,000 designs, CoNN-AL achieves R-squared values between 0.967 and 0.982 using only 2.3% of the full training set, outperforming random sampling and approaching the performance of models trained on much larger datasets.
Related stories
arXiv Statistics ML
Oct 7
Multi-Task Active Learning with Efficient Resource Allocation
arXiv:2610.07045v1 Announce Type: cross Abstract: Many scientific studies allow costly auxiliary information to be collected during data labeling but not at deployment. Examples include diagnostic te...
arXiv Computer Vision
Sep 10
One Loop, Two Gains: Can Active Learning win the Lottery for Free?
arXiv:2609.10311v1 Announce Type: cross Abstract: The lottery ticket hypothesis posits the existence of winning tickets: sparse subnetworks that, when trained in isolation from their original initial...
arXiv Machine Learning
Aug 21
CarBench: A Comprehensive Benchmark for Neural Surrogates on High-Fidelity 3D Car Aerodynamics
arXiv:2512. 07847v2 Announce Type: replace Abstract: Benchmarking has been the cornerstone of progress in computer vision, natural language processing, and the broader deep learning domain, driving algorithmic innovation through standardized datasets and reproducible evaluation protocols.
arXiv AI
Sep 18
Green-ELM: Efficient Analytic Learning via High-Dimensional Random Projections
arXiv:2604. 15613v4 Announce Type: replace-cross Abstract: We present Green-ELM, a non-iterative neural architecture that replaces gradient-based optimization of the output layer with a closed-form analytic solution over a fixed, high-dimensional random feature representation.
arXiv Machine Learning
Oct 8
Cross-Domain Pretraining for Steady-State Neural CFD Surrogates
The paper presents cross‑domain pretraining for neural surrogates in steady‑state computational fluid dynamics (CFD). By training models across varied geometries, boundary conditions, and fidelities, the authors show that such pretraining improves zero‑ and few‑shot performance compared to training from scratch or transferring from domain‑specific experts. Finetuning a pretrained cross‑domain model can reduce errors by 2–3× at the same sample size and require eight times fewer samples to reach the same error, with benefits that grow with model size and dataset diversity.
arXiv AI
Aug 5
One Knob to Rule Them All: A Unified Optimal Transport View of Cold-Start Active Learning
arXiv:2608. 03249v1 Announce Type: new Abstract: Cold-Start Active Learning (CSAL) aims to select a valuable subset from an unlabeled pool without any prior knowledge or human assistance.
arXiv AI
Sep 7
Partial Inverse Design of High-Performance Concrete Using Cooperative Neural Networks for Constraint-Aware Mix Generation
The paper introduces a cooperative neural network framework for partially inverse designing high‑performance concrete (HPC) mixes. It combines an imputation model with a surrogate strength predictor and is trained cooperatively, enabling it to produce valid, performance‑consistent mix designs in a single forward pass without retraining for different constraints. Compared to baseline methods, the approach achieves higher strength consistency (R² 0.84–0.89) and reduces mean squared error by 42–60%.
arXiv Machine Learning
Oct 8
Edge Accuracy Is Not Enough: Why Dynamics-Learned Structure Fails to Transfer to Inverse Problems
The paper investigates why transferring relational structure learned from abundant forward-simulation data fails for inverse problems, despite theoretical expectations. It demonstrates that dynamics‑learned graphs, such as those from Neural Relational Inference, can worsen performance by over 100% compared to a task‑optimised attention baseline, and that this failure persists across multiple evidence lines. The authors formalise the gap as an approximation‑error issue and propose a lightweight transferability test based on Jaccard similarity to predict successful transfer.
arXiv Machine Learning
Sep 22
Physics-residual machine learning predicts oxygen-evolution catalyst activity beyond the training range from sparse polarization measurements
arXiv:2609.23549v1 Announce Type: new Abstract: Screening oxygen-evolution catalysts on combinatorial libraries requires deciding which candidates receive the remaining measurements. The deciding act...
arXiv Machine Learning
Sep 30
Meta-learning accelerates detector design optimization
arXiv:2609.35827v1 Announce Type: cross Abstract: The quality of a detector design is ultimately determined by the quality of the inference it enables, that is, by the accuracy with which the quantit...
arXiv Statistics ML
Oct 1
Mitigating Representation Gaps in Amortized Bayesian Inference with Auxiliary Supervision
arXiv:2609.39525v1 Announce Type: new Abstract: Casting Bayesian inference as a neural network optimization problem targeting an amortized posterior is attractive, as it extends to otherwise intracta...