✓✓ Beats tuned baseline
2026
Add a geometry-guided infill operator to a population optimizer used for black-box neural-network tuning. Fit a local Jacobian from recent parameter perturbations and validation-residual vectors, generate a damped Gauss-Newton candidate for exploitation, and sample exploratory candidates in the same Jacobian-derived metric. The host optimizer retains selection, population survival, covariance adaptation, and its total evaluation budget; only a configurable fraction of new candidates is replaced…
Useful6/10
Difficulty5/10
Novelty7/10
✗ Mechanism failed
2026
Replace AdamW or SGD updates on simplex-valued routing probabilities with a logarithmic-barrier mirror step. The update remains strictly positive, avoids projection-induced zero coordinates, and can approach a boundary solution asymptotically while retaining the paper's theoretically motivated O(log k/k) convex convergence behavior.
Useful6/10
Difficulty5/10
Novelty6/10
Unverified
2026
Treat the number K of minibatches between expensive control updates as a review period: the controlled neural dynamics use parameters or decisions computed at time nK and hold them fixed until (n+1)K. Scan K, estimate first and second finite differences of validation loss or episodic return, and use the resulting nonmonotone-to-convex or concave phase diagram to select an update frequency rather than assuming that more frequent updates are always better.
Useful5/10
Difficulty4/10
Novelty6/10
Unverified
2026
Monitor optimizer convergence over a cycle of p updates instead of judging every update independently. Estimate the p-step contraction factor and effective convergence order from parameter or loss errors, then reduce learning rate only when the cycle-level contraction worsens, avoiding false alarms caused by alternating or oscillatory iterates.
Useful5/10
Difficulty4/10
Novelty7/10
Unverified
2026
Replace the direct nonlinear loss step by a scalar-auxiliary-variable discretization of a gradient flow. The optimizer maintains an auxiliary value representing the square root of the nonlinear energy, so the coupled update has a discrete modified-energy decrease even when the step size is not restricted by the local curvature of the loss.
Useful5/10
Difficulty6/10
Novelty7/10
Unverified
2026
Replace the explicit parameter update \(\theta_{k+1}=\theta_k-\eta\nabla L(\theta_k)\) with an approximate generalized proximal step defined by a simple map \(v\). The map is chosen so that the gradient operator and v satisfy an empirical pair-monotonicity condition, allowing larger stable outer steps and reducing oscillations in stiff or highly curved neural-network training.
Useful5/10
Difficulty7/10
Novelty6/10
Unverified
2026
Replace an empirically chosen momentum or learning-rate modulation by a forcing amplitude calibrated to the homoclinic energy balance of a reduced optimizer mode. The controller deliberately operates below the separatrix-crossing threshold when stable refinement is desired, or slightly above it when the optimizer must escape a basin. This creates a falsifiable transition prediction rather than merely adding noise or tuning a schedule.
Useful5/10
Difficulty6/10
Novelty8/10
Unverified
2026
Periodically project a rectangular neural-network weight matrix onto an approximately orthonormal-column matrix using LU-preconditioned CholeskyQR rather than ordinary QR or a polar iteration. Pivoted LU handles badly scaled and nearly dependent columns, while Householder orthogonalization of the LU factor produces a triangular preconditioner that makes the subsequent Cholesky step safer in fp16 or bfloat16.
Useful5/10
Difficulty6/10
Novelty5/10
Unverified
2026
Use a two-gradient predictor-corrector average as the gradient supplied to Adam, retaining trajectory smoothing while avoiding the three or four gradient evaluations required by full RK3. Vary the mixing coefficient to test whether the reported regularization comes from gradient averaging itself rather than from high-order integration.
Useful5/10
Difficulty4/10
Novelty5/10
Unverified
2026
Treat a scalar training control, such as task-mixture weight, weight decay, or sparsity penalty, as a parameter ramped through a sharp optimization transition. If the model starts from a highly correlated pretrained or partially trained state, compensate for the predicted marginal logarithmic memory by slowing the ramp according to a fitted logarithmic factor rather than using a pure power-law schedule.
Useful5/10
Difficulty5/10
Novelty8/10
Unverified
2026
For a recurrent, state-space, implicit, or complex-valued neural network, partition the local input-output Jacobian into amplitude and phase channels and penalize excessive sensitivity in either channel. This transfers the paper's voltage-source stiffness mechanism to feature magnitude and phase, producing a stability monitor that can distinguish harmless amplitude sensitivity from destructive phase rotation.
Useful5/10
Difficulty5/10
Novelty6/10
Unverified
2026
Represent selected activations, weights, or optimizer accumulators as four floating-point limbs and evaluate products through tensor-core matrix multiplications encoding limb convolution. Retain the convolution components during reductions and renormalize only at block boundaries, avoiding branch-heavy multi-double arithmetic inside every multiply-add.
Useful5/10
Difficulty6/10
Novelty6/10
Unverified
2026
Construct a block of gradient, preconditioned-gradient, or Hessian-vector-product directions without performing full-dimensional Gram-Schmidt. Use a random sketch to orthogonalize the block cheaply, then solve a small generalized eigenproblem using the true parameter-space overlap matrix so the extracted curvature modes are accurate for the generated subspace. Use the selected curvature modes to form a damped or trust-region optimizer step.
Useful5/10
Difficulty6/10
Novelty7/10