Every idea extracted from recent arXiv mathematics papers — verified and unverified. Click an idea to open its full card; badges show the empirical verdict.
Construct overlapping attention windows from a block design instead of using one dense sequence-by-sequence attention matrix. Every token appears in exactly $r$ windows and every token pair co-occurs in exactly $\lambda$ windows, giving uniform coverage and avoiding the uneven connectivity of arbitrary sparse masks.
Compress a transformer KV cache by selecting actual past tokens whose key or hidden-state columns form a stable basis for all cached tokens. Instead of retaining tokens with the largest attention scores or leverage scores independently, compute rank-revealing pivoting of the leading right-singular-vector matrix and retain its pivot columns, then evaluate attention using the representatives plus an optional low-cost residual correction.
Train a matrix-valued neural layer under an exact or near-exact Stiefel constraint while using an l1 or row-group sparsity penalty. During early training, use manifold proximal-gradient steps to identify a stable nonzero support; once the support stops changing, switch to Newton-CG steps restricted to the smooth intersection of the Stiefel tangent space and the fixed-support subspace. This can reduce the number of optimizer iterations needed to obtain sparse, well-conditioned projections.
Replace dense query-key attention with an adaptive cross approximation constructed from selected query and key pivot tokens. At each rank, choose the pivot pair that removes large estimated residual energy, update the residual by a rank-1 cross correction, and stop when the residual estimate reaches a target tolerance. The resulting factorization computes approximate attention using a small number of landmark interactions while adapting to the actual token distribution.
Replace dense self-attention by a sparse mask on a one-dimensional or ordered token geometry, retaining local neighbors and adding long-range edges with probability proportional to distance raised to \(-(1+\sigma)\). Use \(\sigma\approx0.8\text{--}0.85\) as the initial regime because the paper finds that this range supports delocalized, GOE-like connectivity despite sparse bonds. The resulting layer has linear or near-linear attention cost while maintaining long-range paths.
Replace ordinary low-precision multiply-add accumulation in selected neural-network reductions with a two-word floating-point accumulator updated by the paper's branch-free DW-FMA network. The high word retains the main sum and the low word stores the rounding residual, improving cancellation behavior without the control-flow divergence of conditional compensated summation.
Partition tokens into a rectangular grid and use Latin-square labels to define several sparse attention heads. Each head attends only to tokens sharing a row, column, or Latin label, while orthogonality guarantees that every pair of labels occurs at most once, reducing systematic blind spots and repeated collisions. The resulting masks are deterministic, reusable across examples, and can be generated without learned routing scores.
Replace quadratic self-attention over a sequence with a bank of K auxiliary exponentially decaying states whose rates are fitted directly from the empirical autocorrelation of the sequence features. Each mode captures a distinct time scale, so the module can represent short- and long-range dependencies with O(TK) computation and O(K) recurrent memory rather than storing all previous tokens. Constrain decay rates to be positive and use the paper's extended Markovian block structure to obtain a…
Use a smoothed Burg entropy as the mirror map in a proximal-gradient optimizer for positive or simplex-valued neural parameters. The optimizer performs a Bregman-proximal step instead of an additive Euclidean update, while the smoothing parameter avoids the singularity of ordinary Burg entropy at zero.
Replace a dense graph-attention or token-mixing matrix by a resolvent-like interaction operator and truncate it to graph neighborhoods whose radius is selected from an estimated spectral gap. Unlike fixed-window sparse attention, the sparsity level is tied to a measurable stability parameter and has an explicit exponential tail criterion.
Construct a sparse attention layer by sampling backward token histories as a continuous-time branching process rather than allowing every query to attend to every key. Each active ancestor either dies or branches into a bounded number of candidate ancestors, with branching probability controlled by a small parameter. The branch-out penalty predicts exponentially small probability of long, highly branching histories, providing a direct knob for receptive-field size and attention FLOPs.
Replace confidence-only masked diffusion decoding with an adaptive scheduler that chooses batches whose unrevealed tokens have low conditional total correlation given the already revealed context. The scheduler should preserve large parallel batches when token predictions are conditionally independent, but split highly dependent tokens into separate rounds to reduce forward-KL error.
For a fixed structured graph, preprocess its adjacency matrix into the linear-time matrix-vector multiplication data structure guaranteed for classes of linear neighborhood complexity. Replace every dense aggregation Y=MX in a GNN by batched queries to this exact data structure, reducing a dense O(n^2d) aggregation to O(nd) after O(n^2) one-time preprocessing. This is especially useful for dense graphs from bounded-clique-width, bounded-expansion, minor-closed, twin-width, or related structured…
Replace a dense coordinate-kernel interaction among N points by an orthogonal samplet transform with a sparse detail-detail matrix and a small polynomial branch. Detail basis vectors have vanishing moments, so smooth low-frequency behavior is represented by a few polynomial coefficients while localized residual interactions become sparse in the transformed domain.
Replace dense token attention with K asynchronous local nonlinear consensus updates on a sparse token graph. Each selected token is moved to the unique incident p-mean of its neighbors, producing a tunable message-passing operator that is averaging for p=2 and nonlinear for other p.
Replace dense aggregation over patches, points, or graph neighbors with a state-dependent positive quadrature rule. A latent encoder produces the current manifold coordinate z, a small controller chooses a subset of entities, and a convex redistribution step assigns nonnegative weights so that the subset preserves the full aggregation on a learned feature basis. The method is especially suitable for neural operators and vision attention where many spatial entities are redundant along the data…
Construct a sparse directed attention mask over ordered tokens by treating token pairs as edges labeled by time or position. Apply a recursive three-hop spanner construction so that every dense temporal route has a sparse route using at most three times as many hops. Run attention only on the selected edges and use three sparse message-passing layers, or a small learned hop-composition block, to approximate dense causal attention.
Replace dense attention weights with a compactly supported anisotropic bump derived from the obstacle solution, using one learnable ellipsoid per attention head or feature group. Tokens outside the learned ellipsoid receive exactly zero weight, while tokens inside receive smoothly decaying weights according to a fractional exponent. The learned positive-definite matrix represents orientation, scale, and correlations between feature dimensions.
Insert a sparse heavy-tailed nonlocal mixing operator into a residual sequence, graph, or spatial network so that information can traverse distant positions without stacking many local layers. Use the critical tail exponent s = 1/2, whose truncated first moment grows logarithmically and predicts an effective propagation distance proportional to depth times log depth rather than merely depth.
Replace a sparse graph layer's separate edge transformations with one joint low-rank factorization of all transformations entering each target node. For target node i, concatenate the neighbor matrices horizontally, project all neighbor features into a shared low-dimensional receiving basis, and reconstruct one output; retain the self transformation exactly. This can reduce edge-parameter storage and message-passing FLOPs when the incoming block row has rapidly decaying singular values.
Replace random node or token subsampling by a subset selected to preserve the averages of low-frequency graph signals exactly or approximately. Build a graph over nodes or tokens, compute its Laplacian eigenspace, and choose a fixed-cardinality subset whose eigenvector moments match those of the full graph; use the retained subset for graph pooling or downstream attention.
Replace fixed-budget token or patch pruning with greedy selection that combines a teacher-derived relevance score and Gaussian-process mutual information. Select an item when it is both relevant and non-redundant, and stop when the largest remaining information gain falls below a calibrated threshold instead of retaining a fixed number of items.
Approximate a dense symmetric interaction matrix in a neural layer by \(\widehat A=C\widehat M C^{\top}\), but compute the small core \(\widehat M\) from a two-sided sketched least-squares fit rather than from the landmark principal submatrix. This preserves signed or indefinite directions and avoids exploding outputs caused by an almost-singular \(A(I,I)\).
Construct a sparse attention variant whose positional interaction kernel is piecewise polynomial rather than an arbitrary softmax score. Store key-value content in exact monomial moments within a spatial or learned-coordinate box hierarchy, then evaluate all queries from those moments without materializing the query-by-key matrix.