CAWI replaces standard random initialization of input-to-hidden weights in randomized neural networks with samples drawn from a data-fitted copula that preserves observed feature dependencies, yielding consistent accuracy gains on 83 classification benchmarks.
Title resolution pending
29 Pith papers cite this work. Polarity classification is still indexing.
29
Pith papers citing it
verdicts
UNVERDICTED 29representative citing papers
Probability-of-Hit acquisition function ranks perturbation candidates by posterior probability of threshold exceedance, with asymptotic optimality proof and up to 6.4% gains on real immunology data.
LE-SAM inverts SAM by fixing the loss budget instead of the parameter-space radius, yielding better generalization across benchmarks.
ResRL decouples shared semantics between positive and negative responses in LLM reinforcement learning via SVD-based projection residuals, outperforming baselines including NSR by up to 9.4% on math reasoning benchmarks.
o1-like models overthink easy tasks; self-training reduces compute use without accuracy loss on GSM8K, MATH500, GPQA, and AIME.
Multi-query attention shares keys and values across heads in Transformers, greatly reducing memory bandwidth for faster decoding with only minor quality loss.
FNO exhibits strong frequency bias with sharp OOD error growth on high-frequency inputs in wave equations, while DeepONet shows milder degradation despite higher baseline error.
TBPO derives a token-level preference optimization objective from sequence-level pairwise data via Bregman divergence ratio matching that generalizes DPO and improves alignment quality.
CTFusion is a live-CTF streaming benchmark that prevents data contamination by forwarding only the first correct flag per challenge under a shared team account.
R-GFM constructs multi-scale Riemannian graph-of-graphs to learn geometry-adaptive representations, reducing structural domain generalization error and delivering up to 49% relative gains on downstream graph tasks.
The general regularization scheme is extended to conditional density estimation, yielding a new estimator with proven convergence rates that matches or beats the Nadaraya-Watson estimator in experiments.
LLMs contain identifiable COCO neurons that enable implicit self-correction against stereotypes; targeted editing of these neurons improves fairness and robustness to jailbreaks while preserving generation quality.
A counterexample disproves the conjecture that minimal filling architectures of polynomial neural networks always have unimodal hidden layer widths.
QueST replaces local point tracking with persistent semantic queries that globally attend to spatio-temporal features and apply 3D grounding to suppress drift, cutting absolute point error by 67.7% versus TAP-Net on long articulated sequences.
NSER uses zero-shot LLMs to induce behavioral rules from RL trajectories, grounds them in differentiable first-order logic, and applies the symbolic structures to dynamically reweight experience replay for better sample efficiency.
FAR-SIGN achieves adversary-resilient fully asynchronous optimization via signed directional projections and two-timescale correction, with almost-sure convergence to stationary points at rates O(n^{-1/4+ε}) first-order and O(n^{-1/6+ε}) zeroth-order.
Kinematics-GS reparameterizes Gaussian shapes along motion trajectories with a kinematic prior to reconstruct dynamic 3D scenes from blurry monocular videos by separating dynamic and static components and using coarse-to-fine optimization.
Future-rhyme information is linearly decodable at line boundaries across model families and strengthens with scale, yet only Gemma-3-27B causally depends on it, with the driver migrating to the boundary around layer 30 and localizing to five attention heads.
A Hessian-free stochastic Runge-Kutta LMC algorithm achieves strong order 1.5 with two gradient evaluations per step and uniform-in-time convergence O(d^{3/2} h^{3/2}) in non-log-concave settings.
NPMixer improves multivariate time series forecasting accuracy by combining a data-adaptive wavelet decomposition with hierarchical neighboring patch mixing via MLPs and channel mixing on high-frequency components.
ExecuTorch is a unified PyTorch-native deployment framework that enables seamless on-device execution of AI models across heterogeneous hardware while preserving original PyTorch semantics.
Unsupervised behavioral mode discovery combined with mutual information rewards enables RL fine-tuning of multimodal generative policies that achieves higher success rates without losing action diversity.
SynerMedGen introduces generation-aligned understanding tasks and a two-stage training strategy that enables strong zero-shot medical image synthesis performance and outperforms specialized models when generation training is added.
A malicious agent in multi-agent LLM consensus systems can be trained via a surrogate world model and RL to reduce consensus rates and prolong disagreement more effectively than direct prompt attacks.
citing papers explorer
-
From Passive Reuse to Active Reasoning: Grounding Large Language Models for Neuro-Symbolic Experience Replay
NSER uses zero-shot LLMs to induce behavioral rules from RL trajectories, grounds them in differentiable first-order logic, and applies the symbolic structures to dynamically reweight experience replay for better sample efficiency.