Fig. 3: Process-aware latent space analysis and sampling.
From: A process-synergistic active learning framework for high-strength Al-Si alloys design
The panels highlight how the latent space guides process-specific sampling and optimization. a Visualization of alloy compositions in the process-aware latent space, where distinct clusters represent different PRs, with data points color-coded according to UTS values. Symbol shapes indicate varying ranges of Si content. b Sampling workflow in the compositional latent space for a specified PR (exemplified by GC + T6). c Progressive optimization through three iterative cycles for GC + T6 PR, showing the latent space evolution and GMM probability density map refinement toward high-strength alloy compositions. d Accelerated optimization pathway for GC + HE PR, achieving high-strength alloy compositions within only a single iteration.
