轻量级优化器FANoS-v2用反馈控制提升训练稳定性，性能优于AdamW。
FANoS-v2: Feedback-Controlled Momentum with Thermostat Damping for Lightweight Neural Optimization
- 引入反馈控制器与温控阻尼，动态调节更新能量。
- 在多个数据集上比AdamW提升2.7个百分点，但耗时增加约50%。
- 适合关注训练稳定性和可复现性的研究者使用。
FANoS 是一个基于 PyTorch 的优化器，通过在 RMS 预条件动量基础上引入标量反馈控制器来调控更新能量。其公开实现将动量存储在参数更新单位中，采用非负温控阻尼系数，支持对角、因子化和原始梯度预条件，并提供用于稳定性审计的诊断信息。本文完整给出了已发布优化器的数学规范，包括精确的参数单位更新、物理更新模式、有界对数比温控、自适应预条件软化、预热保护机制以及实验性快速配置。报告 v0.2 的证据：在五种子集的简样本 MNIST、Fashion-MNIST 与 CIFAR-10 实验中，Fast 配置相较于 AdamW 平均提升 0.889、2.197 和 2.666 个百分点，但壁钟时间分别高出 49.8%、61.6% 和 56.8%。初步的科学计算、PINN 与 EEG 测试结果不一致，仅作为假设生成。证据表明，FANoS 为研究阶段的 α 版优化器，具备可复现的轻量视觉信号，但也存在明确的运行瓶颈。
原文摘要 · Abstract (English)
\FANOS{} is a PyTorch optimizer that augments RMS-preconditioned momentum with a scalar feedback controller over update energy. The public reference implementation stores momentum in parameter-update units, applies a non-negative thermostat damping coefficient, supports diagonal, factored, and raw-gradient preconditioning, and exposes diagnostics intended for stability audits. This study gives a complete mathematical specification of the released optimizer, including the exact parameter-unit update, the study-equation physical update mode, bounded log-ratio thermostat control, adaptive preconditioner softening, warmup guardrails, and the experimental \Fast{} profile. We report the v0.2 evidence: five-seed reduced-sample MNIST, Fashion-MNIST, and CIFAR-10 experiments show mean top-1 gains of 0.889, 2.197, and 2.666 percentage points over AdamW for \Fast{}, but with 49.8\%, 61.6\%, and 56.8\% higher wall-clock time. Preliminary scientific, PINN, and EEG smoke tests are mixed and are treated as hypothesis-generating only. The evidence supports \FANOS{} as an alpha-stage research optimizer with a reproducible lightweight-vision signal and an explicit runtime bottleneck.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。