Every idea extracted from recent arXiv mathematics papers — verified and unverified. Click an idea to open its full card; badges show the empirical verdict.
Train the raw neural policy to anticipate the safety shield instead of relying on it indefinitely. Add the expected process-correction distance to the PPO objective and adapt its multiplier with a primal-dual update so that policy correction remains below a specified budget.
Replace a binary action mask with a lossless probability-preserving shield for sequential neural decisions. At state s, retain only actions that preserve current and future feasibility, then transfer the raw policy mass of excluded actions to admissible frontier actions according to process distance and the actor's existing safe-action preference.
Use discovered infinitesimal generators to create small continuous transformations of hidden states and force a neural predictor to commute with those transformations. This converts symmetry discovery into self-supervised augmentation without prespecifying a group, canonical coordinates, or hand-designed equivariant layers.
Augment a latent neural state-space model with an observable-coordinate residual that is first learned flexibly and then projected onto a constrained library of interpretable coupling terms. Train or collect data only after checking that the trajectory sufficiently excites the candidate terms; this prevents a latent model from fitting arbitrary hidden-state effects that are unidentifiable from the observations.
Modify the flow-matching denoising velocity at every integration step by the smallest Euclidean correction that satisfies a control-barrier inequality on the entire generated action chunk. Aggregate waypoint and constraint barriers with a smooth minimum, so the sampler remains differentiable and can enforce safety without retraining the action model or applying a discontinuous final-action projection.
Add a discrete-event safety shield between a partially observed neural policy and the environment. The policy proposes a forcing action, but the shield permits it only when the same decision is safe for every latent plant state compatible with the current observation; otherwise it returns a certified inconsistency or a conservative fallback. This converts forcing consistency into an implementable robust action-selection rule rather than trusting a single estimated hidden state.
Use the observation Jacobian to remove from a neural latent dynamics model the component of its drift that is locally inconsistent with the observed manifold. Apply this projected drift only to generate particle proposals, and retain exact importance-ratio correction so that proposal projection improves particle coverage without changing the target posterior.
Augment a neural latent or sequence model with a Gaussian behavior head that predicts an entire future trajectory jointly from the observed prefix and planned inputs. Instead of recursively applying only a point predictor, condition the learned joint trajectory covariance on the available prefix, producing a corrected future mean and uncertainty that incorporates temporal correlations.
Estimate the local contraction rate along minibatch couplings of neural ODE or flow-matching trajectories instead of using one global Lipschitz lower bound. Use the resulting displacement-weighted rate to trigger adaptive solver tolerances, training-time regularization, or early stopping when the transported distributions have entered a strongly contracting region.
Augment a learned neural state-space model with an online regularized least-squares confidence set for its local linearization or last-layer dynamics, then propagate a homothetic uncertainty tube around every predicted trajectory. Use the tube to tighten RL action constraints, reject unsafe imagined rollouts, or weight training examples by certified prediction reliability. The mechanism should improve long-horizon behavior specifically when model uncertainty is large, rather than acting as an…
Add a slowly updated adversarial sampler over training contexts, domain shifts, perturbation levels, or task instances. The neural network trains normally on samples from the current mixture, while a contextual bandit increases probability on contexts with high recent validation loss or catastrophic constraint violation. Unlike static domain randomization, this curriculum explicitly targets current failure modes without changing the model architecture.
Treat a scalar projection of the stochastic training trajectory as a generalized current and use a finite-time concentration bound to decide when its mean estimate is reliable. Increase batch size, reduce the learning rate, or stop collecting samples when the bound predicts that the probability of a misleading gradient estimate is below a target confidence level.
Use an online estimate of the loss barrier separating the current basin from candidate neighboring basins to tune optimizer noise or a trust-region radius. The paper predicts that the current- or power-maximizing barrier is nonzero and approximately matched to an effective harmonic-mean temperature, U_0^* approximately equal to T_act, providing a concrete schedule for increasing or decreasing exploration.
Add a bounded probing perturbation to the inputs or intermediate outputs of a neural sensor-fusion model, and choose the perturbation by maximizing separation between the predicted trusted-output set and output sets induced by candidate sensor attacks. Bounded feature and measurement uncertainty are propagated through local neural Jacobians as zonotopes, giving a conservative, geometry-based exposure objective rather than relying on random noise. Training can use the resulting margin as a…
Partition neural-network parameters into blocks and update each block using a stochastic proximal best response, followed by Krasnoselskii relaxation. The relaxation factor and minibatch size become explicit stability knobs: aggressive stochastic updates are damped, while larger batches are used when the estimated update variance approaches the mean-square stability boundary.
Wrap stochastic optimization or iterative neural inference in a controller that measures how far the state distribution moves during each interval and compares this motion with the available noise-dependent entropy-production budget. The controller increases the learning rate or reduces inference steps only while the trajectory remains inside the predicted speed-limit region, preventing fast jumps that cause accuracy collapse.
Wrap a neural policy with an online disturbance estimator and a zonotopic reachability shield. Instead of rejecting actions using a permanently worst-case disturbance set, update the disturbance zonotope from observed transition residuals and accept an action only when the resulting reachable set remains inside the safe region.
Build a recurrent or state-space model with a base state carrying task-relevant dynamics and an explicitly contracting auxiliary state. If the training loss or energy depends on the auxiliary state, replace it by a quotient loss plus an analytically known telescoping correction; long-run optimization and invariant averages are then unchanged, while transient fiber effects decay geometrically.
Replace a conventional linear decoder in an autoencoder or latent state-space model with an explicit quadratic manifold decoder, allowing a small latent vector to represent curved and transport-like state trajectories. Add a dynamics-aware invariance loss that penalizes the discrepancy between the time derivative of the quadratic manifold and the neural dynamics evaluated on that manifold.
Build a continuous-time RNN or neural state-space model whose latent dynamics possess two stable periodic attractors representing persistent sequence modes, then inject weak calibrated noise to induce rare transitions between them. Instead of treating mode switching as an arbitrary classifier event, estimate the minimum transition action and tune the noise level or an explicit control input so that the observed switching rate matches the desired rate. This should improve long-horizon multimodal…
Attach a deterministic supervisory automaton to a neural policy or sequence model and mask every event disabled by the current supervisor state. Use a short receding-horizon planner over admissible events to resolve conflicts between neural preferences and shared-resource constraints. The network scores useful actions, while the automaton supplies an exact safety layer.
Replace a fixed-size random transition gate with a risk-calibrated gate whose test count is chosen from the estimated probability of a critical event and the cost of shipping a model that misses it. The gate should combine ordinary i.i.d. rollouts with planner-generated probes aimed at high-cost boundaries, because uniform sampling can make a dangerous model appear perfectly accurate.
Add an exact backward-conditioning module to a neural state-space model so trajectories satisfy a terminal label, target set, initial-state restriction, or prescribed event count without rejection. The module computes a backward feasibility message and reweights each neural transition toward states that can still satisfy the constraint, producing a conditioned process equivalent to a Doob transform. For large latent spaces, the exact message can be approximated by a value network and its…
Train a neural value or latent-dynamics model with temporal-difference targets before enforcing a stiff differential-equation residual, and ramp the physics weight only after the critic has become predictive. For a stochastic dynamical model, the residual is computed using the infinitesimal generator, while terminal, safe, and failure boundary conditions are imposed through separate penalties.