2581 valuable arXiv mathematics papers the pipeline has analyzed — each with a short note on what its math gives to machine learning.
The paper provides a tractable mechanism by which increasing L2 regularization causes a cascade of feature-detection phase transitions in deep linear networks. Under whitened inputs and approximately balanced, singular-vector-aligned…
The paper provides a concrete framework for reducing communication in distributed optimization while controlling the optimization error introduced by infrequent synchronization and compressed client updates. Its most transferable…
The paper identifies a sharp regime boundary for robust average-reward learning: the relevant comparison is between the target accuracy ε and the perturbation scale σH₀. Its useful transferable asset is not the tabular sample bound…
The paper provides a mathematically controlled way to replace an all-pairs interaction field by interactions inside freshly randomized batches. The transferable asset is not the Biot–Savart kernel itself, but the unbiased…
The paper's transferable contribution is an active-set geometry for systems defined by pairwise non-overlap inequalities: realised contacts determine both the local feasible cone and the equality subspace of contact-preserving motions…
The paper's transferable asset is an incremental persistence computation scheme: instead of recomputing zigzag persistence after every change, it maintains homological decompositions while blocks split, merge, or change direction, using…
The paper identifies a concrete universality class for noisy, number-conserving operator propagation: Green-function weight behaves like a directed wave in a random medium, with KPZ exponents rather than ordinary diffusive exponents. The…
The paper combines an invertible neural surrogate with conformal prediction to certify feedforward tracking of an unknown nonlinear dynamical system. Its transferable mechanism is to replace a difficult nonconvex inversion problem with an…
The paper gives a discrete Gauss–Bonnet construction for a graph viewed through its simplicial structure: a vertex curvature is computed entirely from simplex counts in its unit sphere, and the curvatures sum exactly to the Euler…
The paper gives an exact algebraic method for extracting simple cycles from weighted graph walks: assign an independent variable to every vertex, take a trace or log-determinant generating series, and retain only squarefree monomials. This…
The paper establishes a universal control-budget theorem for the mean first-passage time of any finite continuous-time Markov chain: the logarithmic sensitivity to each individual transition rate has magnitude at most one, while the…
The paper gives a sharp worst-case upper-tail bound for sums of independent nonnegative variables with bounded means, identifying sparse Bernoulli spikes at the threshold as the extremal distribution. The transferable asset is a…
The transferable asset is a principled state-dependent taming of potentially superlinear subgradients, combined with Langevin noise, rather than a heuristic fixed gradient clip or a smoothing approximation. This can become a nonsmooth…
The paper's transferable contribution is a control principle for transport dynamics: localized damping need not act everywhere or at every instant if every relevant trajectory accumulates a uniform amount of damping over a finite time…
The paper constructs highly nonnormal finite-dimensional generators whose forward semigroups are uniformly bounded and exponentially decaying, while inverse dynamics exhibit large amplification. The transferable lesson is that…
The paper provides a matrix-level optimization primitive whose exact solution is the negative polar factor of a momentum matrix. Unlike coordinate-wise updates, this direction solves a spectral-norm constrained linear minimization and…
The paper gives a sharp, dimension-dependent comparison between probing a bipartite matrix with arbitrary global observables and probing it only with product observables. The transferable asset is a certified approximation principle…
The paper introduces a canonical multiscale descriptor for interactions among colored point clouds: at radii t_1,...,t_k, it records the Euler characteristic of the intersection of the corresponding unions of balls. The transferable asset…
The paper provides a constructive way to learn hidden dynamics when training data contain only marginal observations rather than complete trajectories or latent states. Its transferable asset is the characteristic formulation: simulate…
The paper provides a concrete inference-time mechanism for binary neural networks: reorder each dot product by descending weight magnitude so that high-impact sign evidence arrives first. The strongest transferable component is a…
The paper provides a verifiable regularity mechanism for conditional expectation operators: regularity of the conditional density implies that the operator maps between prescribed function spaces and can be bounded or Hilbert–Schmidt…
The paper provides an exact, rejection-free sampler for valid correlation matrices using row-normalized Bartlett factors: independent Gaussian off-diagonal entries and chi-squared diagonal entries are normalized row by row to produce an…
The paper contributes a nonstandard constraint for conditional generative emulators: the latent noise mean should map to the conditional mean of the next physical state. This anchors an otherwise stochastic generator at a predictable…
The paper's transferable asset is the Multiple Scalar Auxiliary Variable (MSAV) treatment of nonlinear energy terms: a difficult nonconvex contribution is represented through scalar variables, allowing each time step to reduce to linear…