Indexed text attachment
operads-modules-machine-learning.tex
Source for Operads, Modules, and Differentiation in Machine Learning
View paper · Download original file
Lines 1061–1160 of 1174. Source text is displayed literally.
1061kernels being compared. In a shared-child computation or a1062program that carries conversation state between branches,1063marginal laws can agree even when joint laws differ.1064For instance, two fair bits can be independent in one1065evaluator and always equal in the other. Their marginal laws1066agree, yet exclusive-or distinguishes the evaluators.1067In that setting a valid comparison has to track the joint law1068of the consumed inputs.1069 1070The bound compares output laws. A ground-truth predicate, a1071formal checker, or a task-specific notion of utility supplies1072a separate test of the answer. A one-colour grammar with a1073constant answer at every operation can be perfectly consistent1074and wrong on every question with a different correct answer.1075Thus tree consistency can diagnose certain difficulties of1076substitution while soundness still depends on the interpretation1077of the elementary steps and on the external correctness test.1078 1079\section{Scope and outlook}1080\label{sec:scope}1081 1082The application to smooth architecture evaluation concerns the degree-zero part of the monograph's setting. The passage to \(\mathbb R[P]\) ensures that the module operators are defined on real chain complexes concentrated in degree zero. It gives a model for organizing derivatives and a check on compositional implementations. The original feature operations may be nonlinear. For a piecewise smooth layer the calculation applies on regions of differentiability. A kink, such as that of ReLU at zero, requires an additional derivative convention.1083 1084There are several different meanings of tangent structure. In the smooth calculation, a tangent space \(T_x X\) consists of directions in a parameter or state space. In the Quillen-theoretic setting, the tangent \(\infty\)-category of a presentable \(\infty\)-category at an object is the stabilization of the slice. For operadic algebras under stable hypotheses, its comparison to operadic modules is a theorem about that stabilization~\cite[Corollary~1.0.5]{harpaz2019}. This supports the role of operadic modules in homotopical linearization. Applying that theorem to learning requires a model of the architecture or data in the specified category, as well as a definition of the perturbations to be measured. The present paper directly establishes the smooth derivative interpretation and computes a separate derived symmetry example.1085 1086The clearest uses of the monograph in machine learning are the interpretation of typed architecture substitutions, the module law for one-input sensitivities, and the derived analysis of symmetry relations when ordinary balancing loses homology. For AI systems whose components return stochastic outputs, agreement between two routes is an empirical question about suitably typed kernels. The consistency calculation in Section~\ref{sec:ai-applications} supplies a diagnostic with clear sampling assumptions. The bridge in both deterministic and stochastic cases comes from the marked-slot module and the precise rule for composing it.1087 1088\paragraph{Reproducibility.} The source bundle contains the full LaTeX and a standard-library Python program for the worked numerical derivative and symmetry checks in Section~\ref{sec:examples}. These checks supplement the proofs and are limited to the worked examples.1089 1090\paragraph{Authorship.} The public byline is Anonymous. A language-model assistant helped draft, typeset, check, and revise the manuscript under the submitter's direction.1091 1092\begin{thebibliography}{12}1093\small1094\setlength{\itemsep}{2pt}1095 1096\bibitem{kriz-may}1097I.~K\v{r}\'{\i}\v{z} and J.~P. May.1098\newblock \emph{Operads, algebras, modules and motives}.1099\newblock Ast\'erisque 233, Soci\'et\'e Math\'ematique de France, 1995.1100\newblock \url{https://www.numdam.org/item/AST_1995__233__1_0/}.1101Page locators in this paper follow the authors' 109-page version,1102\url{https://math.uchicago.edu/~may/BOOKS/KM.pdf}.1103 1104\bibitem{fong2019}1105B. Fong, D.~I. Spivak, and R. Tuy\'eras.1106\newblock Backprop as functor: A compositional perspective on supervised learning.1107\newblock In \emph{Proc. 34th Annual ACM/IEEE Symposium on Logic in Computer Science}, 2019.1108\newblock \url{https://arxiv.org/abs/1711.10455}.1109 1110\bibitem{cruttwell2022}1111G.~S.~H. Cruttwell, B. Gavranovi\'c, N. Ghani, P. Wilson, and F. Zanasi.1112\newblock Categorical foundations of gradient-based learning.1113\newblock In \emph{Programming Languages and Systems}, LNCS 13240, pp. 1--28, 2022.1114\newblock \url{https://arxiv.org/abs/2103.01931}.1115 1116\bibitem{gavranovic2024}1117B. Gavranovi\'c, P. Lessard, A. Dudzik, T. von Glehn,1118J. G. M. Ara\'ujo, and P. Veli\v{c}kovi\'c.1119\newblock Position: Categorical deep learning is an algebraic theory of all architectures.1120\newblock In \emph{Proc. 41st International Conference on Machine Learning},1121PMLR 235, pp. 15209--15241, 2024.1122\newblock \url{https://proceedings.mlr.press/v235/gavranovic24a.html}.1123 1124\bibitem{vagner2015}1125D. Vagner, D.~I. Spivak, and E. Lerman.1126\newblock Algebras of open dynamical systems on the operad of wiring diagrams.1127\newblock \emph{Theory and Applications of Categories}, 30(51):1793--1822, 2015.1128\newblock \url{https://arxiv.org/abs/1408.1598}.1129 1130\bibitem{harpaz2019}1131Y. Harpaz, J. Nuiten, and M. Prasma.1132\newblock Tangent categories of algebras over operads.1133\newblock \emph{Israel Journal of Mathematics}, 234:691--742, 2019.1134\newblock \url{https://arxiv.org/abs/1612.02607}.1135 1136\bibitem{zaheer2017}1137M. Zaheer, S. Kottur, S. Ravanbakhsh, B. P\'oczos,1138R. Salakhutdinov, and A.~J. Smola.1139\newblock Deep Sets.1140\newblock \emph{Advances in Neural Information Processing Systems} 30, 2017.1141\newblock \url{https://arxiv.org/abs/1703.06114}.1142 1143\bibitem{spivak2026}1144D.~I. Spivak.1145\newblock Compositional dynamics in learning and mechanics.1146\newblock arXiv preprint, 2026.1147\newblock \url{https://arxiv.org/abs/2606.28984}.1148 1149\bibitem{bottman2026}1150N. Bottman and K. Richardson.1151\newblock Operads for compositional reasoning in LLMs.1152\newblock arXiv preprint, 2026.1153\newblock \url{https://arxiv.org/abs/2606.13634}.1154 1155\bibitem{bottmanliu2026}1156N. Bottman, Y. Liu, and K. Richardson.1157\newblock Operadic consistency: A label-free signal for compositional reasoning failures in LLMs.1158\newblock arXiv preprint, 2026.1159\newblock \url{https://arxiv.org/abs/2606.13649}.1160