Every idea extracted from recent arXiv mathematics papers — verified and unverified. Click an idea to open its full card; badges show the empirical verdict.
Replace Cox-de Boor evaluation of each cubic B-spline edge activation with its fixed truncated-power expansion. Normalize each scalar edge input to a bounded knot coordinate, evaluate the five shifted cubic positive-part terms in parallel, and contract them with the learned spline coefficients inside one fused kernel.
Estimate the largest certified input perturbation radius for a neural network using nested reduced primal and dual linear programs rather than solving the complete verification LP immediately. The primal sequence gives certified feasible robustness reserves, while the dual sequence gives valid upper bounds; verification may stop as soon as the interval width is below a prescribed tolerance.
Construct a residual network from independently attachable modules, but permit only a number of modules whose aggregate feedback gain lies inside a delay-dependent admissible interval. Estimate deployed end-to-end latency and each module's local Jacobian gain, then reject or bypass additional modules when the predicted delayed-loop stability boundary is crossed. This turns variable-width or depth scaling into a falsifiable control problem rather than an empirical choice.
Replace the global EMA update for each linear-layer momentum matrix with a delta-rule update that learns the current output-side gradient value only along the current input-key direction. Frequently occurring directions are corrected repeatedly, while rarely visited directions are not unnecessarily overwritten or uniformly decayed. Use the resulting matrix as the ordinary momentum buffer in SGD, AdamW, or another optimizer.
Replace black-box differentiation through an embedded LP decision with an analytic Jacobian computed from the LP’s active basis. A neural policy emits LP coefficients or right-hand sides; the LP returns the decision, while the backward pass uses the basis inverse and dual sensitivity, avoiding solver unrolling and finite-difference noise.
Turn an iterative optimization or equilibrium computation inside a neural network into a differentiable layer whose backward pass solves the implicit adjoint system with conjugate gradients or GMRES using only automatic-differentiation matrix-vector products. This avoids storing unrolled iterations and avoids explicit Hessian or Jacobian construction, enabling longer solver horizons and lower-memory implicit architectures.
Replace a square dense projection in a Transformer or MLP with a trainable recursive butterfly matrix. The layer preserves multiscale channel interactions while constraining every complementary row-column block to rank at most k, reducing parameters and enabling recursive structured matrix-vector products. Unlike an arbitrary sparse layer, the construction has an explicit recursive factorization and a quasi-optimal approximation guarantee among matrices with the same butterfly rank.
Replace a dense neural-network weight tensor with a graph tensor network whose physical modes and internal edge ranks are specified by a sparse rank-adjacency matrix. Unlike tensor-train or hierarchical Tucker layers, the graph can contain selected cycles and skip connections between tensor modes, allowing the factorization topology to match correlations in the weight tensor. Fit the layer with GTN-SVD at a prescribed tolerance and compare accuracy, parameter count, and tensor-contraction…
Before quantizing a matrix product, reparameterize its factors as A'=AT and B'=T^{-1}B, preserving the exact full-precision product while changing the quantization difficulty of each factor. Choose a positive diagonal T=diag(t_1,...,t_K) that minimizes predicted post-quantization product error, rather than using output-channel scaling or a fixed heuristic grid. The gauge can be shared across several products when transformed-copy cost matters.
Replace uniform collocation for a fixed random-feature neural PDE solver with sampling from the leverage-score density of the operator-applied features. Whiten the retained residual feature space before solving for output coefficients, so the sampled least-squares matrix has an identity-like expected Gram rather than inheriting severe anisotropy from the differential operator. The same construction can be used for a linearized neural network by treating Jacobian features as the trial functions.
Replace Euclidean or entrywise Kronecker fitting of a layer curvature matrix with its affine-invariant projection onto G = A tensor B. Use the resulting factors as a compact SPD preconditioner in the optimizer, while solving the projection through logarithmic residual partial traces and Armijo line search.
Compress each hidden layer by retaining directions that are simultaneously reachable from the observed input distribution and observable at the network output. Unlike PCA or SVD, the retained subspace is weighted by downstream task sensitivity, so high-variance but output-irrelevant directions can be removed while low-variance predictive directions are preserved.
Replace a full neural-network Gauss–Newton solve with a damped solve in an adaptively constructed low-dimensional parameter subspace. The subspace contains the current gradient, recent accepted updates, Krylov curvature directions, and randomized Jacobian-curvature probes, and is enlarged whenever its projected gradient fails to capture enough descent information.
Represent a large linear map acting on a Cartesian 3D grid and multiple physical channels as a TT-matrix, while retaining separate TT blocks for channel couplings that have different semantics. Apply the layer by sequential contractions with TT cores rather than materializing a dense matrix or a full 3D convolution kernel. Rank truncation provides an explicit accuracy-versus-memory knob and can be applied after optimizer updates.
For a neural network with a trainable linear head or low-rank adapter, store feature vectors from recent minibatches and select a finite set that is sufficiently independent. Apply Modified Gram-Schmidt to obtain orthonormalized memory directions, then add residual corrections along these directions so the local parameter-error dynamics have an identity coefficient matrix rather than a poorly conditioned empirical Gramian. The method predicts a sharp transition after the buffer first contains…
Replace a dense Haar or Gaussian random projection with a streamed product of random two-coordinate rotations followed by coordinate subsampling. The transform is exactly orthogonal before subsampling, requires only a list of rotation triples, and the paper's pseudo-mixing result predicts that degree-two statistics relevant to norm preservation and Johnson–Lindenstrauss embeddings become Haar-like after only O(n polylog(n)) rotations.
Insert a linear Johnson–Lindenstrauss bottleneck around a set of jointly processed representations, choosing its width from the sharp finite-set dimension bound rather than from the model's nominal hidden size. The projection should preserve pairwise distances between tokens, patches, or retrieved items, allowing a downstream attention or MLP block to operate at lower width while retaining the geometry relevant to similarity computations.
Build a neural operator around explicit input and output measurement spaces rather than forcing the network to consume and emit a fixed grid. The same learned latent surrogate can be reused on alternative sensor layouts or query meshes through reconstruction and re-encoding maps, with a consistency loss enforcing agreement between measurement pipelines.
Attach a polyharmonic spline decoder to a coordinate MLP or use it as a standalone neural-field output head over a large set of spatial anchors. The decoder represents the output as a low-degree polynomial trend plus a PHS kernel expansion, while FMM evaluates all anchor-to-query interactions in approximately linear or near-linear cost. When coefficients must be fitted or periodically recalibrated, solve the constrained interpolation system with projected conjugate gradients and a sparse…
Use the lifted second-moment operator to adapt the residual step size of a deep residual network or neural ODE under multiplicative layer noise. Instead of choosing a fixed residual coefficient, shrink or enlarge it online to keep the predicted covariance-growth factor below a target margin, producing a stochastic stability controller for depth and inference time.
Replace fixed-length binary dot products with accumulations whose terms are processed in descending order of weight magnitude. Stop as soon as the current partial sum is larger in magnitude than the total absolute magnitude of all remaining terms; the output sign is then guaranteed to equal the full dot-product sign, eliminating unnecessary additions without changing accuracy.
Replace the standard diagonal or identity preconditioner used when solving an implicit neural layer with a coarse/fine Schur-complement preconditioner. The hidden state is decomposed into a low-dimensional coarse subspace and its orthogonal complement; the coarse interaction is solved accurately, while the fine block receives a damped approximate inverse. The method is especially suitable for deep equilibrium models, implicit MLPs, and Newton or quasi-Newton training of residual dynamics.
Compress an existing dense neural-network weight matrix into a recursive butterfly operator using Gaussian sketches of complementary blocks. This is useful for deployment or distillation: the dense model provides an oracle for matrix-vector products, while the compressed model stores only recursive transfer bases and small cores. The generalized Nyström identity gives exact reconstruction for rank-k blocks and a principled approximation route for numerically low-rank blocks.
Replace the exact matrix-polar normalization in Muon with the smoothed feedback \(h_\epsilon(M)=M(M^\top M+\epsilon I)^{-1/2}\). This retains singular-vector-aware updates and approximately unit-normalizes dominant spectral modes, but avoids unstable behavior when the momentum matrix is rank deficient or has tiny singular values.