发表机构
Cranberry-Lemon University; University of the Witwatersrand（蔓越莓柠檬大学; 威特沃特斯兰德大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
研究针对自改进学习系统的非平稳目标问题，提出协同进化框架PRAXIS，通过理论证明其稳定性保证，实验验证其在三类任务上的有效性。
AI 中文摘要
自改进学习系统会调整数据选择、优化及辅助符号组件，产生超出标准学习假设的非平稳目标。我们提出PRAXIS，一种协同进化框架，将生成器、学习器和符号归档建模为相互作用的动力学过程。我们证明，KL约束的生成器更新与受控的归档权重移动可限制单步目标漂移；在次高斯噪声下，归档更新能使程序相对于任何固定比较器获得持久的累积效用优势；随机梯度下降可实现平均平稳性保证，其退化由累积目标漂移决定。在视觉鲁棒性、关系图推理和算法图推理上的实验显示，生成器稳定、学习器损失降低，且归档浓度与这些理论机制一致。
英文摘要
Self-improving learning systems adapt data selection, optimization, and auxiliary symbolic components, inducing nonstationary objectives outside standard learning assumptions. We introduce \textsc{PRAXIS}, a co-evolutionary framework that models generators, learners, and symbolic archives as interacting dynamical processes. We prove that KL-constrained generator updates and controlled archive-weight movement bound one-step objective drift, that archive updates suppress a program relative to any fixed comparator with a persistent cumulative utility advantage under sub-Gaussian noise, and that stochastic gradient descent achieves an average-stationarity guarantee whose degradation is governed by cumulative objective drift. Experiments across visual robustness, relational graph reasoning, and algorithmic graph reasoning exhibit generator stabilization, decreasing learner loss, and archive concentration consistent with these theoretical mechanisms.
