✗ Failed on benchmark
2026
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.
Useful7/10
Difficulty5/10
Novelty5/10
△ Mechanism confirmed, baseline not beaten
2026
Estimate an expensive fine-model trace or quadratic-form quantity using a telescoping sum over cheap-to-expensive neural approximations. Allocate many probes to cheap levels and only a few probes to the expensive level, exploiting strong correlation between adjacent levels to reduce variance at fixed compute. Candidate levels include truncated Transformer depth, reduced width, low-rank curvature, coarser graph resolution, or progressively tighter implicit-solver tolerances.
Useful7/10
Difficulty6/10
Novelty6/10
△ Mechanism confirmed, baseline not beaten
2026
Replace independent Hutchinson vectors used to estimate traces of neural-network curvature operators with graph-coloring probing vectors. Coordinates that are far apart in an interaction graph share a color, so one probe simultaneously covers many coordinates while reducing variance from localized off-diagonal matrix entries. Apply this to Hessian-trace regularization, Fisher-trace diagnostics, or layerwise curvature estimates used by adaptive optimizers.
Useful7/10
Difficulty5/10
Novelty7/10
✓✓ Beats tuned baseline
2026
Use the differentiable covariance chart to construct a Fisher-information preconditioner for the edge and innovation parameters of a linear-Gaussian neural module. Instead of applying an isotropic Euclidean update, whiten parameter steps according to how strongly they change the predicted Gaussian distribution. This targets ill-conditioning caused by redundant paths, correlated latent nodes, and badly scaled innovation covariances.
Useful7/10
Difficulty6/10
Novelty5/10
✗ Failed on benchmark
2026
Replace the sign-flip-only dynamics of high-index saddle search with low-rank inverse-curvature scaling on the estimated negative-curvature subspace. Directions with small negative Hessian eigenvalues then receive approximately curvature-independent updates instead of extremely slow updates proportional to their tiny curvature.
Useful7/10
Difficulty6/10
Novelty7/10
△ Mechanism confirmed, baseline not beaten
2026
Construct a positive learning-rate schedule offline by minimizing the worst residual of every prefix on a normalized curvature interval, rather than optimizing only the final training horizon. The schedule is evaluated through the exact quadratic residual polynomial p_n(lambda) = product_{k=1}^n (1 - eta_k lambda), so every prefix is constrained to make progress across multiple curvatures.
Useful7/10
Difficulty5/10
Novelty6/10
✓✓ Beats tuned baseline
2026
Replace full-precision all-reduce parameter averaging in synchronous distributed training with the paper's compressed gradient-tracking recursion. Each worker maintains a model state, a gradient-tracker state, and two communication memories; only compressed differences from the memories are exchanged, while the tracker preserves the global-gradient increment despite compression.
Useful7/10
Difficulty6/10
Novelty5/10
✓ Mechanism works
2026
Search for a compact symbolic optimizer instead of selecting among fixed AdamW-like formulas. Encode optimizer programs as token sequences, learn a continuous variational representation of those sequences, and use a Gaussian-process Bayesian optimizer to propose promising update rules based on short neural-network training rollouts.
Useful7/10
Difficulty6/10
Novelty6/10
✗ Mechanism failed
2026
Use the activation singular-value spectrum to allocate a fixed zeroth-order parameter budget across layers instead of assigning the same rank everywhere. Layers with a large discarded singular-value tail receive more coefficient directions, while spectrally compressible layers use smaller adapters, preserving the gradient-relevant subspace under a global memory and query budget.
Useful7/10
Difficulty4/10
Novelty6/10
✗ Mechanism failed
2026
Train a network to predict the context-dependent observation matrix rather than the latent inverse parameters themselves, then compute the latent parameters with a differentiable ridge-regression solve. This gives one model that can assimilate arbitrary observation vectors, exposes the conditioning of the inverse problem, and avoids forcing an MLP to learn the entire map from observations to parameters.
Useful7/10
Difficulty5/10
Novelty7/10
✗ Mechanism failed
2026
Replace conventional nested bilevel optimization with simultaneous primal-dual updates that enforce inner-model stationarity through a Lagrange multiplier. Add quadratic dual regularization and projection onto a bounded ball, while estimating all Hessian-vector terms using finite differences of ordinary gradients.
Useful7/10
Difficulty5/10
Novelty6/10
✗ Mechanism failed
2026
Replace ordinary momentum with a semi-implicit velocity update containing viscous damping and a proximal dry-friction step, while evaluating the gradient at a look-ahead parameter point. The dry-friction proximal operator exactly zeros sufficiently small velocities, which may suppress late-training oscillations and create finite-time stationarity instead of merely asymptotic velocity decay.
Useful7/10
Difficulty4/10
Novelty7/10
✗ Mechanism failed
2026
Use Adam normally, but periodically estimate the spectrum of the Adam-preconditioned Hessian and add a damped low-rank Newton correction when the preconditioned curvature is strongly ill-conditioned or the gradient is concentrated in flat directions. The correction is computed only in a small Lanczos subspace, so the method targets cross-coupled ill-conditioning without materializing or inverting the full Hessian.
Useful7/10
Difficulty6/10
Novelty6/10
Unverified
2026
Construct a generative or recurrent neural architecture with several contractive or mildly expanding branches, and explicitly control the geometric complexity of its invariant set using the sub-additive singular-value pressure of branch-Jacobian products. Instead of regularizing only the operator norm, the model can preserve anisotropic directions while targeting a desired attractor dimension, potentially improving coverage of structured data without uncontrolled folding or collapse.
Useful6/10
Difficulty6/10
Novelty8/10
Unverified
2026
Add a low-dimensional, trusted sentinel state to the optimizer or recurrent inference controller. The sentinel is driven by a secret probe and a protected gain, so unauthorized gain changes produce a predictable shift in its equilibrium even when the main neural dynamics remain numerically stable. Monitor the estimated equilibrium and trigger rollback or quarantine when the measured shift exceeds the expected noise envelope.
Useful6/10
Difficulty4/10
Novelty8/10
Unverified
2026
Replace fixed-beta RMSprop with a bias-corrected, slowly increasing-beta schedule. The paper’s error decomposition contains a stochastic-approximation term of order gamma_n and a memory term of order (1-beta)^2; setting (1-beta_n)^2 proportional to the current learning rate makes the memory contribution decay at the same scale as the optimization error instead of leaving a fixed residual.
Useful6/10
Difficulty4/10
Novelty5/10
Unverified
2026
Replace decentralized parameter averaging with consensus on cumulative local gradient states, followed by a barrier-FTRL update that stays strictly inside a convex feasible set. This is particularly suitable for federated learning with heterogeneous clients and for simplex-constrained mixture, router, or adapter parameters, where Euclidean projection can be unstable or expensive.
Useful6/10
Difficulty5/10
Novelty6/10
Unverified
2026
Replace a single global preconditioner for a multi-penalty neural objective with additive corrections adapted to the joint kernels of the penalty Jacobians. The optimizer is designed to remain effective when individual penalty weights change independently, avoiding the severe conditioning degradation that occurs when a correction space misses a singleton or partial joint kernel.
Useful6/10
Difficulty6/10
Novelty6/10
Unverified
2026
Train a neural state-space model whose one-step dynamics are linear in a fixed analytic feature vector, and use the empirical feature Gram matrix to detect whether passive trajectories identify the dynamics. Add data collection or replay only when the Gram matrix is poorly conditioned; the analytic-feature assumption predicts that persistent excitation should emerge without deliberately visiting every operating mode.
Useful6/10
Difficulty5/10
Novelty7/10
Unverified
2026
Replace ordinary parameter updates for a constrained neural network with an annealed reciprocal-manifold flow. Each differentiable inequality constraint remains strictly satisfied during the optimization trajectory, avoiding projection or a per-step quadratic program. This is most useful for safety-critical policy learning, bounded network outputs, parameter-budget constraints, or training with explicit robustness inequalities.
Useful6/10
Difficulty5/10
Novelty5/10
Unverified
2026
Use two points with approximately equal minibatch loss to construct an ellipcenter: the intersection of the normal lines through the two points, where the normals are their gradients. The resulting update uses local curvature information in the span of two gradients and can be relaxed toward the current parameters or combined with momentum.
Useful6/10
Difficulty5/10
Novelty7/10
Unverified
2026
Extend regularized extragradient with proximal operators so nonsmooth penalties such as group sparsity, nuclear norms, or parameter constraints are applied at both prediction and correction stages. This can produce sparse or low-rank adapters while retaining the look-ahead stabilization for the smooth inner residual.
Useful6/10
Difficulty5/10
Novelty7/10
Unverified
2026
Use a decaying Tikhonov term to make inner training dynamics select a stable outer-preferred solution, and evaluate the regularized operator at a look-ahead point before updating parameters. This is intended for convex heads, adapters, equilibrium layers, or locally monotone inner objectives rather than unrestricted nonconvex end-to-end training.
Useful6/10
Difficulty5/10
Novelty6/10
Unverified
2026
Add a deliberately nonconservative, antisymmetric parameter-space force to ordinary gradient descent, with its amplitude controlled by an empirically estimated stability margin. The force should move parameters around elongated loss valleys instead of repeatedly descending and stopping along the same local gradient direction, while damping preserves convergence. The method directly tests whether nonzero circulation can improve traversal of flat or ill-conditioned regions without destabilizing…
Useful6/10
Difficulty5/10
Novelty8/10