✗ Failed on benchmark
2026
Add a training-time regularizer that keeps the empirical joint covariance of hidden activations on multiple inputs close to the recursively predicted NNGP covariance. The regularizer targets the finite-width fluctuations quantified by the Wasserstein result, and is particularly appropriate for recurrent networks and attention blocks with shared weights, where hidden states at different positions or time steps are statistically coupled.
Useful6/10
Difficulty5/10
Novelty6/10
✗ Mechanism failed
2026
Turn row dropout into an adversarial conditioning problem rather than independent Bernoulli noise. At each training step, search for a subset of surviving channels or measurements with unusually small least singular value, train the downstream network on that subset, and gradually increase the search strength so training directly exposes failure modes hidden by average-case dropout.
Useful6/10
Difficulty6/10
Novelty8/10
✗ Failed on benchmark
2026
Regularize the end-to-end Jacobian singular-value distribution of a deep network toward the explicit free small-loss law generated by independently mixed projection-like layers. The target controls several gradient-spectrum moments, including the predicted fraction of nearly preserved directions, instead of controlling only the average gradient norm.
Useful6/10
Difficulty6/10
Novelty7/10
✗ Mechanism failed
2026
Replace the first dense layer on q-ary categorical features by a Fourier interaction layer containing only monomials whose coordinate support is at most s. Use a Bohnenblust–Hille-inspired quasi-norm on coefficients, separately for each interaction order, to prevent a small number of high-order interactions from dominating the output. The resulting model has an explicit interaction-order knob and can be tested against a dense MLP at matched parameter count.
Useful6/10
Difficulty5/10
Novelty7/10
✗ Mechanism failed
2026
Replace an unconstrained scalar activation or nonnegative gate with a finite positive mixture of rational Bernstein basis functions. The learned function is monotone and concave on the nonnegative half-line, while its derivatives have controlled alternating signs; this can prevent pathological feature amplification and gives an interpretable shape prior. Use the paper's sharp exponent restriction τ≤1/2 rather than treating the power as an arbitrary hyperparameter.
Useful6/10
Difficulty4/10
Novelty7/10
✓✓ Beats tuned baseline
2026
Choose gradient clipping thresholds from an explicit worst-case tail probability implied by an observed kurtosis bound, rather than using a fixed norm threshold or an empirical percentile. For a standardized centered gradient coordinate, the threshold achieving target outlier probability \(\delta\) is obtained by analytically inverting the paper's sharp tail formula.
Useful6/10
Difficulty4/10
Novelty5/10
✗ Failed on benchmark
2026
Replace repeated full-curvature factorizations when evaluating many Bayesian neural-network hyperparameter settings with one shared low-rank approximation of the data-induced precision update. Reuse the same curvature subspace while changing prior precision, observation-noise precision, or layerwise regularization, evaluating each candidate through a small rank-r matrix and Woodbury solves.
Useful6/10
Difficulty5/10
Novelty5/10
✓✓ Beats tuned baseline
2026
Replace a continuous scalar latent or probability with a stochastic count Y generated by Y|X=x ~ Binomial(n,x), and feed Y/n to the downstream network. Regularize the aggregate count distribution toward the beta-binomial distribution induced by the arcsine input X~Beta(1/2,1/2), while maximizing the mutual information carried by the count. This creates a compact discrete representation with an analytically specified, nonuniform prior that places more mass near the extreme counts without…
Useful6/10
Difficulty4/10
Novelty7/10
✗ Mechanism failed
2026
Replace an unconstrained softmax gate over a finite set of neural experts with exponential weights whose temperature is chosen to satisfy the paper's explicit stability condition. The goal is to prevent low-temperature expert collapse while retaining the model-selection rate when the expert losses are bounded and strongly convex in the prediction.
Useful6/10
Difficulty4/10
Novelty3/10
✗ Mechanism failed
2026
Use the divergence's data-processing principle as a consistency objective between predictions before and after a stochastic augmentation or feature bottleneck. Penalize disagreement under transformations while retaining the asymmetric power-law weighting of the r-deformed divergence.
Useful6/10
Difficulty4/10
Novelty5/10
△ Mechanism confirmed, baseline not beaten
2026
Replace cross-entropy or ordinary Renyi loss between a target distribution and a model distribution with the paper's r-deformed alpha-z divergence. The deformation parameter r provides a controllable power-law alternative to the logarithm, allowing experiments that emphasize hard, low-probability target events differently from standard log losses.
Useful6/10
Difficulty3/10
Novelty6/10
✗ Mechanism failed
2026
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.
Useful6/10
Difficulty4/10
Novelty6/10
✗ Mechanism failed
2026
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.
Useful6/10
Difficulty3/10
Novelty6/10
✗ Mechanism failed
2026
Replace an unconstrained high-order polynomial interaction module with features generated by Gaussian matrix contractions and their exact Wick expansion. The resulting interactions are sums of products of power-sum invariants, with coefficients fixed by perfect-matching counts, providing a low-parameter inductive bias for permutation- or orthogonal-structured data.
Useful6/10
Difficulty5/10
Novelty7/10
Audited (legacy)
2026
Train a classifier or encoder to distinguish shared latent corruption from fresh per-view noise instead of treating repeated observations as conditionally independent given the target. A single persistent state corrupts all views, while each view may additionally receive independent observation noise; the fusion loss marginalizes the persistent state exactly. This should reduce overconfident predictions from repeated but systematically biased augmentations, sensor readings, or retrieved…
Useful6/10
Difficulty4/10
Novelty7/10
△ Mechanism confirmed, baseline not beaten
2026
Add a directional persistent cross-entropy loss between teacher and student activation persistence diagrams. The loss assigns high probability to teacher topological events that the student reproduces, while accumulating the probability of unmatched teacher events in an explicit unexplained-event mass. This penalizes missing teacher structure without requiring teacher and student diagrams to have the same number of points.
Useful6/10
Difficulty7/10
Novelty7/10
✗ Failed on benchmark
2026
Add the paper's local Sine_beta fusion law as an analytic score prior for diffusion models that generate unordered point configurations. The model is trained to match both the usual diffusion score and an explicit short-range repulsion score, including the second-order correction that describes finite-scale fused configurations.
Useful6/10
Difficulty5/10
Novelty6/10
✓ Mechanism works
2026
Replace disjoint-pair estimates of embedding covariance moments with a complete U-statistic over every distinct pair in a minibatch. For embeddings z, the degree-two kernel h(z_i,z_j)=(z_i^T z_j)^2 estimates the spectral moment tr(M^2), where M=E[zz^T]; complete symmetrization reduces the degenerate component of estimator variance from O(1/B) to O(1/B^2).
Useful6/10
Difficulty4/10
Novelty5/10
Unverified
2026
Add asynchronous nonlinear consensus steps between local-SGD updates in a federated or decentralized system. Instead of averaging a participating client with all peers, replace one client parameter vector by the coordinatewise p-mean of neighboring replicas, allowing sparse communication and reduced sensitivity to atypical client models.
Useful5/10
Difficulty5/10
Novelty7/10
Unverified
2026
Replace a raw pairwise repulsion or similarity penalty on neural embeddings by a centered mean-field energy plus a Carleman–Fredholm determinant correction for finite-batch quadratic fluctuations. The correction uses the spectrum of a centered learned kernel, retaining the effect of important fluctuation directions while removing the first-order mean-field component. A Nyström or landmark approximation makes the method practical without forming a large batch-by-batch determinant.
Useful5/10
Difficulty5/10
Novelty7/10
Unverified
2026
Replace ad hoc Gaussian or Laplace noise injection with a Laplace majorant calibrated to the observed finite-range sub-Gamma parameters of a neural perturbation. For convex perturbation losses, the calibration guarantees that the expected loss under the scaled Laplace noise upper-bounds the expected loss under every centered random perturbation satisfying the same Bernstein-type MGF constraint.
Useful5/10
Difficulty6/10
Novelty6/10
Unverified
2026
Replace a learned linear operator by Monte Carlo averages of unitary-conjugated copies, where the conjugators are discrete time-frequency shifts sampled from a narrow phase-space distribution. The conjugation preserves the singular spectrum of each sampled operator, while the expectation penalizes sensitivity to small translations and modulations without directly shrinking the operator norm.
Useful5/10
Difficulty5/10
Novelty7/10
Unverified
2026
Represent augmentation centers or training examples in a low-dimensional torus and accumulate the geometric overlap of their augmentation neighborhoods. Add a finite-horizon penalty that detects latent locations with insufficient accumulated coverage, while constraining center uniformity so that coverage optimization does not collapse all samples to one location.
Useful5/10
Difficulty5/10
Novelty7/10
Unverified
2026
Use the reciprocal arrangement as a probe of whether a learned representation has the intended angular response, and penalize deviations from the paper's universal beta distribution. This converts the theorem into a distribution-level regularizer rather than assuming that the reciprocal layer itself improves task loss.
Useful5/10
Difficulty4/10
Novelty7/10