2581 valuable arXiv mathematics papers the pipeline has analyzed — each with a short note on what its math gives to machine learning.
The paper introduces a spectrum-aware ordering of graph edges using the local Frobenius cosine between an edge's rank-one Laplacian perturbation and the existing Laplacian. This inexpensive proxy identifies edges whose removal aligns…
The paper provides a transferable mechanism linking network degree heterogeneity to nonequilibrium bifurcations in feedback-driven Glauber dynamics. Its key quantitative control parameter is the degree-moment ratio \(\alpha=\langle…
The paper provides a concrete topological tokenization mechanism: convert a graph filtration into a short sequence of interlevel Betti numbers and feed that sequence to a Transformer. The transferable asset is not merely the use of…
The paper offers a transferable multiscale mechanism: occupation-dependent hopping produces density-dependent transport while interactions dephase coherent modes, yielding a crossover between coherent short-scale propagation and incoherent…
The paper offers a closed-loop data-collection mechanism: learn a state-wise predictor of future rollout failure, use it to replace redundant nominal scenarios with failure-prone ones, and apply importance weights so the training objective…
The paper supplies an order-aware optimal-transport construction for measuring how far one distribution is from stochastic dominance, with a parameter γ that interpolates between exact dominance and a weak expectation-based preorder. The…
This paper gives a constructive stochastic-dynamics mechanism for preventing recurrent message passing from collapsing node states: inject independent Gaussian perturbations after every propagation step rather than relying on a one-time…
The paper develops a constructive fixed-charge dynamic program whose value function remains discretely convex after binomial thinning, leading to an (s,S) policy: incur a fixed activation cost only when the resource level is below a…
The paper gives a constructive method for selecting a small set of graph leaders that breaks topological equivalence classes, rather than relying on random or degree-based anchors. Its transferable asset is the combination of…
The paper offers a constructive offset-compensation mechanism: identify both an absolute predictor and an incremental predictor, then use the absolute predictor's steady-state residual as a direct signature of persistent actuator mismatch…
The paper gives constructive decompositions of finite multiparameter modules into interval-supported components under explicit transition-map structure, rather than treating each parameter value independently. This suggests neural…
The paper offers a constructive projection-regularization mechanism for data-driven prediction under errors-in-variables, together with a finite-sample confidence radius that remains valid under temporally correlated closed-loop data. Its…
The paper derives an effective Kardar–Parisi–Zhang (KPZ) phase dynamics by eliminating fast density and reservoir modes from a driven-dissipative condensate, leaving diffusion, a nonequilibrium gradient-square nonlinearity, and white…
The paper supplies a concrete topological-geometric identity that turns an Euler Characteristic Transform into an exact perimeter measurement for any compact tame planar shape. The important transferable asset is that subtracting the Euler…
The paper identifies a concrete anomalous-relaxation mechanism: a preparation that is farther from steady state can relax faster when its projection onto the slowest relaxation mode is smaller, even if its initial free-energy distance is…
The survey identifies data-driven formal verification mechanisms that construct finite abstractions, functional certificates, and compositional guarantees from finite trajectory data rather than an explicit dynamical model. The most…
The paper develops a lattice-and-coset description of divisibility obstructions for packing copies of a fixed graph into dense induced subgraphs. Its transferable asset is not the specific F-factor theorem, but the representation of…
The paper offers a nonstandard structural-controllability construction for directed networks with singular, asymmetric matrix couplings: decompose the matrix-weight space into scalar layers, then bound controllability using layer-specific…
The paper provides a physically grounded construction for turning a capacitance network into a spatially localized, tunable mixing operator. Its transferable asset is the inverse-capacitance coupling K = C^{-1}C_U, whose locality and rank…
The paper combines discrete-gradient chain rules with scalar auxiliary variables (SAV) so that an energy law is enforced exactly at the discrete time-step level while retaining inexpensive linear solves. This is transferable to neural…
The paper provides a constructive method for propagating conservative second-moment tubes through nonlinear stochastic dynamics without discarding Taylor remainders. Its transferable asset is the combination of a uniform nonlinear…
The paper provides a constructive nonequilibrium mechanism: a pair of overdamped degrees of freedom coupled by unequal action-reaction coefficients reaches a stationary state with a nonzero probability current, even under one thermal bath…
The paper supplies an explicit sequence-to-sequence upper bound in which each learning-rate step is weighted by its future learning-rate tail and by the second moment of the stochastic gradient. This exposes a concrete control variable…
The paper gives a sharp finite-sample bias law for the plug-in Shannon entropy estimator when the underlying categorical probabilities have a regularly varying heavy tail. This is transferable to sampled-policy entropy bonuses…