Attention variants, state-space and recurrent cells, normalization and token-mixing schemes — each tested against the standard block it replaces.
Replace dense attention over structured object tokens with attention over role-filler tensor-product representations. A learned query specifies both a role and a filler, retrieves objects matching that binding, extracts a target role, and rebinds the extracted filler into an output object.
Replace the usual random or elementwise-positive linear-attention feature map with a rank-one positive-semidefinite feature map derived from query and key vectors. For normalized inputs, the resulting kernel is the squared inner product, which is nonnegative and gives a geometrically structured interference pattern that is better suited to associative recall than an arbitrary low-rank feature map.
Replace ordinary row-softmax attention with a doubly stochastic Sinkhorn attention plan W, and periodically recover a gauge-fixed pairwise cost from W using the exact double-centering identity. Use this recovered cost to initialize or regularize a structured attention score, making the attention geometry identifiable despite arbitrary query and key row and column offsets.
Augment a 1D neural operator or transformer with explicit tokens for detected discontinuities. Advance each front analytically using the local Rankine–Hugoniot speed and train the network only to reconstruct smooth regions and the residual caused by source terms and grid resolution.
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.
Replace a wide collection of interchangeable near-zero branches with a module whose output is explicitly a quadratic form in the branch-weight Gram matrix. The module preserves the paper's leading-order behavior while making the relevant collective variable explicit and allowing low-rank parameterizations.
Parameterize an entropic OT cost only in directions that can change the transport plan, removing row-plus-column potential directions that are invisible because of OT gauge invariance. Whiten the remaining feature coordinates using their empirical covariance, producing an OT layer whose identifiable parameters have substantially more uniform sensitivity.
Use predicted covariance reduction as a differentiable gate for selecting tokens, views, sensors, or retrieved demonstrations. The gate favors inputs with high expected information gain while accounting for acquisition cost, turning attention and data collection into active observability optimization.
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.
Add a causal memory branch whose lag-response function is represented by a Bernstein polynomial with coefficients constrained to produce a nonnegative, decreasing, convex kernel. The branch aggregates past hidden states using this kernel, giving the model a learnable long-memory profile while preventing oscillatory, negative, or increasing historical influence.
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 ordinary pairwise attention similarity by an affinity averaged over transformed keys or values. The resulting attention is invariant to the group action on either input and avoids requiring the network to learn identical attention patterns for every rotated or transformed copy.
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…
Replace dense coarse-to-fine cross-attention at multiresolution interfaces with a sparse, nonnegative overlap operator whose weighted feature average is exactly conserved between the two resolutions. Use this operator as a low-order path and blend it with an unrestricted neural cross-attention path through a convex limiter that keeps features inside a box or simplex domain. The construction is especially suitable for adaptive token grids, hierarchical graph neural networks, neural operators…
Construct a spatiotemporal neural block from localized functions of a learned parabolic operator instead of unrestricted attention or convolution. Use one filter for fine-scale diffusion and another for coarse-scale temporal aggregation, with the scale ratio controlling information propagation. The block should suppress distant interactions while still permitting long-range mixing through coarse filters.
Use separate measurement-domain and target-domain token sets so a network can infer a field on one spatial domain from sparse observations on another in one forward pass. The same decoder can answer arbitrary target query points, avoiding an optimization loop for each inverse instance.
Replace or augment geometric attention on spatial or point-cloud tokens with a positive fractional kernel containing the paper's inverse-square origin factor. This gives tokens near a designated singular center a controlled increase in receptive-field influence while preserving a scale-invariant distance decay, which may help models represent cusp-like fields, radial singularities, and multiscale spatial interactions.
Construct a sparse token-interaction graph from attention affinities and recursively split it by removing the closed neighborhoods of a small set of separator tokens. Separator tokens retain global communication, while the resulting anti-adjacent components perform local attention independently, reducing quadratic attention and KV-cache costs. The induced Erdos-Pósa theorem supplies a structural diagnostic: graphs with few anti-adjacent long-cycle or theta packings should admit small…
Replace coordinate-wise mean pooling of metric-valued items with a finite representation of their free integral. Each item x in a pointed metric space M is represented through evaluations of learned Lipschitz probes, and the pooled feature is the weighted integral of those probe values. A dual Lipschitz critic estimates the free-space norm of differences between pooled groups, making the representation sensitive to metric geometry while remaining permutation-invariant.