Every idea extracted from recent arXiv mathematics papers — verified and unverified. Click an idea to open its full card; badges show the empirical verdict.
Regularize an intermediate neural representation according to its estimated low-dimensional separability capacity instead of its ambient feature width. Learn feature gates or subspace assignments, estimate the union of active supports, and penalize representations whose Cover capacity exceeds a task-dependent target.
Train a neural field to represent a sphere-valued phase or feature map with a prescribed codimension-two defect set. Add a fractional Sobolev energy to suppress high-frequency oscillations, but enforce topology through a discrete Jacobian or winding-current loss so that smoothing cannot remove holes, filaments, or vortex defects.
Track the implicit l2 regularization induced by adversarial SGD and explicitly correct it when the optimizer drifts toward an undesirable ridge strength. Apply the correction first to the final linear head or a low-dimensional adapter, where feature covariance and ridge estimates are tractable.
Add a differentiable hypergraph layer that converts invariant edge-length features into triangle areas or higher-dimensional simplex volumes before message passing. Select or weight simplices according to the singular values of the length-to-volume Jacobian, so the network receives geometrically independent features rather than many redundant or nearly degenerate measurements.
Insert a scalar flux-correction-style limiter after a neural operator predicts a conservative state or residual. Interpolate between a known-admissible baseline state and the learned high-order candidate, choosing the largest coefficient that satisfies a geometric family of linear inequalities encoding positive density, positive pressure, and subluminal velocity. This retains as much of the neural prediction as possible instead of independently clipping physical variables.
Regularize a classifier on binary or categorical-product inputs with the minimum-norm discrete flow whose divergence matches the model's cube Laplacian. Unlike a direct edge-sensitivity penalty, the flow can route mass nonlocally and combine coordinate changes through an L2 norm, potentially preserving useful interactions while suppressing unstable decision boundaries. The regularizer should be applied to logits or probabilities and combined with the supervised loss, not used alone.
Replace a free-form order-dependent gate with a positive mixture of Mellin powers $(1+s)^{-a}$. This gives a small, interpretable module whose response across the order variable is automatically generated by a positive measure and therefore inherits complete monotonicity, log-convexity, and Hankel-moment structure.
Regularize a neural network's response along an ordered variable by requiring its sampled values to form a positive Hankel moment sequence. This upgrades ordinary pairwise monotonicity or log-convexity penalties into simultaneous constraints on several higher-order interactions, while remaining differentiable and inexpensive for small Hankel order.
Attach predictive distributions to successive information-update steps of a recurrent, state-space, iterative, or diffusion model and penalize violations of the measure-valued martingale condition. The model may become more certain as information arrives, but its later forecasts must not exhibit systematic conditional bias relative to earlier forecasts.
Replace a flat stochastic MoE router by a two-level continuous-time routing model: experts within a group communicate rapidly, while transitions between groups occur slowly. Use the effective class-level stationary distribution as a soft load-balancing prior, reducing routing oscillation while preserving expert specialization.
Add a two-sided cone-restricted spectral penalty to a recurrent or state-space model. Instead of estimating growth using a symmetric singular-value surrogate, jointly optimize a positive right vector and positive left vector in the extended quotient from the paper, targeting a real generalized eigenvalue of the learned non-selfadjoint transition operator.
Parameterize a relative-position or lag-decay function as a finite positive mixture of exponentials instead of learning arbitrary attention bias values. The resulting kernel is completely monotone on positive distances, so it is nonnegative, decreasing, and has alternating derivative signs; the mixture provides several learned memory scales without allowing oscillatory or unstable long-range biases.
Use a learned asymmetric Finsler-like cost instead of the symmetric Euclidean distance in attention logits. The metric has a Riemannian quadratic part and a directional drift term, while a differentiable barrier enforces the strong-convexity condition derived for the paper's extended $(\alpha,\beta)$-metrics. This lets each attention head prefer one direction in feature space without producing pathological, non-convex distance landscapes.
Replace pointwise spectral normalization of an RNN transition with a stability constraint on the entire family of input-conditioned matrices. Use a learned positive-definite metric P so every transition contracts in the same state geometry, approximating the paper's uniform exponential stability and input-forgetting guarantee.
Replace spectral-radius-only stabilization of a recurrent or state-space transition matrix with a numerical-range constraint. Penalize directions in which the Hermitian part of a rotated transition matrix has a large maximal eigenvalue, controlling nonnormal transient amplification and polynomial state propagation.
Replace Euclidean covariance matching with a discrepancy that identifies covariance matrices differing only by per-channel positive rescaling. Apply it to minibatch feature covariances in a representation-alignment, domain-adaptation, style-transfer, or multi-view objective so that the network is penalized for changing correlation structure but not arbitrary channel units.
Add a regularizer that penalizes expensive component births and merges in an embedding when points are partitioned into multiple colors, such as classes, modalities, or augmentation identities. Unlike ordinary contrastive learning, it encourages local regions to contain all required colors and uses the full merge hierarchy rather than only selected positive and negative pairs.
Replace the usual softmax router or soft one-hot penalty with a vector-valued phase-field regularizer whose low-energy states are exactly the expert one-hot vectors. Component-wise barriers create stable categorical phases, while a weaker coupling term suppresses invalid states such as the all-zero vector or multi-expert activation; annealing \(\varepsilon\) produces increasingly discrete routing.
Add a Gaussian KL-UOT-inspired covariance discrepancy to a neural representation loss, using ridge-logdet terms that remain finite when minibatch covariance matrices are rank deficient. Set the unbalanced penalty to \(\tau=\kappa p\), where \(p\) is the feature dimension and \(\kappa\) is tuned over a small logarithmic grid, rather than using a dimension-independent covariance penalty. This directly tests the paper's claim that high-dimensional sample-covariance noise has a critical penalty…
Store a convex object as a direction-indexed vertex tuple and implement composition of objects through componentwise Minkowski addition and nonnegative scaling. This creates a structured residual or compositional layer where convexification is nonexpansive, making perturbation amplification controllable and avoiding repeated generic geometric optimization.
Add a pseudo-determinant-based connectivity objective to a neural model that predicts graph edge weights, attention adjacency, or sparse routing links. Maximizing the Laplacian pseudo-determinant rewards many globally distributed spanning trees, discouraging disconnected or bottlenecked learned graphs without requiring a discrete connectivity constraint.
Use the paper's q-ary overlap inequality as a regularizer for categorical neural networks. Two independently sampled attention, routing, or message-passing supports should rarely overlap in many locations; penalizing the moment q^{|S\cap S'|} discourages redundant histories and correlated interference between heads or experts.
Given arbitrary pairwise preference logits, project their skew-symmetric part onto the additive-consistent subspace before converting logits into probabilities or rankings. This removes cyclic inconsistency using the Frobenius-nearest consistent matrix, guaranteeing transitive pairwise predictions while preserving the closest possible signal under squared error.
Replace random Fourier features or a dense sinusoidal positional encoding with a compact bank whose frequencies are the continued-fraction denominators of an irrational number. Inverse-frequency amplitudes provide multiscale structure with a controlled sub-Lipschitz regularity profile, while lacunarity reduces the number of frequencies needed to represent oscillatory structure.