从像素到光子单元（PCells）：一种用于光子组件创建的神经符号方法
From Pixels to PCells: A Neurosymbolic Approach to Photonic Component Creation
浏览论文内容
中文总结 AI 辅助
该研究提出神经符号系统PixCell，用多模态智能体将视觉呈现的光子组件转为参数化程序，经验证器引导修正的模型在光子组件设计任务上IoU显著提升，建立了相关设计的受控框架。
中文摘要 AI 辅助
我们提出了PixCell，一种神经符号系统，其中多模态智能体将视觉呈现的光子组件转换为小型领域特定语言（DSL）中基于几何基元的参数化程序。该系统支持确定性视觉验证，使得评估的成本比生成尝试的成本更低。虽然使用多种子采样和迭代修正的模型仅达到0.416的平均最佳回合IoU，但通过PixCell接口和验证器的多模态智能体始终超过0.9的平均IoU，在8个组件目标上分别达到0.974和0.955的分数，同时满足源合约。这些结果表明，前沿多模态智能体能够可靠地从视觉目标理解并生成可执行的参数化表示。使用这些实时参数，对干涉仪的跨栈研究重构出满足8.0 nm自由光谱范围目标和220-nm SOI、400-nm SiN、400-nm TFLN建模堆栈上原始 footprint 约束的基元程序。PixCell还将基于论文的分束器从视觉重构通过SOI全波模拟，产生对称传播和平衡输出。最后，相同的可执行验证器提供训练奖励和数据集，用于在无监督演示的情况下，通过LoRA和GRPO训练Qwen3.6-35B-A3B模型。在8个排除训练的论文图上，其平均冠军IoU从初始8次尝试后的0.422提升到3次验证器引导修正回合后的0.491。因此，这些结果建立了一个用于测量、重定向和改进视觉到参数化光子组件设计的受控框架。
英文摘要
We present PixCell, a neurosymbolic system in which multimodal agents convert a visually presented photonic component into a parametric program over a small domain-specific language (DSL) of geometric primitives. A system enabling deterministic visual verification renders evaluation asymmetrically cheaper than the generation attempt. While models using multi-seed sampling and iterative revision reach a mean best-turn IoU of only 0.416, multimodal agents through PixCell's interface and verifier consistently exceed 0.9 mean IoU, with scores reaching 0.974 and 0.955 across eight component targets while also satisfying source contracts. These results demonstrate that frontier multimodal agents can reliably understand and render executable parametric representations from visual targets. Using these live parameters, cross-stack studies on an interferometer reconstruct primitive programs that satisfy an 8.0 nm free spectral range target and the original footprint constraint on modeled 220-nm SOI, 400-nm SiN, and 400-nm TFLN stacks. PixCell further carries a paper-derived splitter from visual reconstruction through SOI full-wave simulation, producing symmetric propagation and balanced outputs. Finally, the same executable verifier supplies a training reward and dataset used to train a Qwen3.6-35B-A3B model with LoRA and GRPO without supervised demonstrations. On eight training-excluded paper figures, its mean champion IoU rises from 0.422 after eight initial attempts to 0.491 after three verifier-guided revision rounds. These results therefore establish a controlled framework for measuring, retargeting, and improving visual-to-parametric photonic component design.
发表机构
- Massachusetts Institute of Technology（麻省理工学院）
- Research Laboratory of Electronics（电子学研究实验室）
- University of Chicago（芝加哥大学）
- The College（学院）
机构由 AI 辅助整理，请以论文原文为准。
