发表机构
NimbusBCI(NimbusBCI)
机构由 AI 辅助整理,请以论文原文为准。AI 中文总结
研究针对冻结脑电图编码器微调难的问题,提出Nimbus Personalizer,通过与主干无关的API,可在异构冻结主干上运行,在多数据集和基础编码器上验证,相比传统方法成本低且能恢复精度增益,结果具探索性。
AI 中文摘要
冻结的脑电图编码器不断增加;每个模型的微调默认设置无法扩展。我们提出了Nimbus Personalizer:一种契约编码,用于连接到BrainState的贝叶斯头部(可选仿射中间层),它位于异构冻结主干上,无需为每个架构设置新的个性化堆栈。贡献在于与主干无关的应用程序编程接口,而不是作为机器学习新奇事物的嵌入上的线性判别分析,因此原始设备制造商只需集成一次并更换主干。证据表明,相同的表面在五个经典主干EEGNet、Shallow、Deep、Conformer、ATCNet x四个运动想象数据集(18个单元)以及同一Personalizer下的基础编码器(REVE)上运行。在存在嵌入能力的情况下,头部是一个廉价的默认中点,与热启动微调或提示学习相比,适应壁时间成本低几个数量级,同时恢复了大部分微调精度增益;仅在干净时进行校准在18个单元中的12个单元中成立。头部增益支持了应用程序编程接口在存在能力的情况下有用的证据。主题级置信区间确定了最清晰的数据集,其他地方跨度为零。所有结果都是探索性的(主题级自举,无验证性测试);我们在控制层的配套工作中讨论了何时升级适应的决策逻辑。
英文摘要
Frozen EEG encoders proliferate; per-model fine-tune defaults do not scale. We present Nimbus Personalizer: one contract encode to Bayesian head to BrainState (optional affine mid-tier) that sits on heterogeneous frozen trunks without a new personalization stack per architecture. Thesis (systems): the contribution is the trunk-agnostic API - not LDA-on-embeddings as an ML novelty - so OEMs integrate once and swap trunks. Evidence: the same surface runs on five classical trunks EEGNet, Shallow, Deep, Conformer, ATCNet x four MI datasets (18 cells) and on a foundation encoder (REVE) under the same Personalizer. Where embedding capacity exists, the head is a cheap default mid-point versus warm-start fine-tune or PEFT, costing orders of magnitude less adaptation wall time while recovering much of the fine-tune accuracy gain; calibration-only-when-clean holds in 12/18 cells. Head gains are supporting evidence that the API is useful where capacity exists. Subject-level confidence intervals identify the clearest dataset and span zero elsewhere. All results are exploratory (subject-level bootstrap, no confirmatory tests); the decision logic for when to escalate adaptation is addressed in our companion work on the control layer.
Comments20 pages, 12 figures. Companion paper on the control layer forthcoming