First population risk bounds for KANs under mini-batch DP-SGD with correlated noise, using a new non-convex optimization analysis combined with stability-based generalization.
Probability inequalities for sums of bounded random variables.Journal of the American Statistical Association, 58(301):13–30
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
citation-role summary
background 1
method 1
citation-polarity summary
years
2026 6verdicts
UNVERDICTED 6representative citing papers
The paper presents randomized tests with explicit query bounds for properties including number of leaves, maximum degree, typical distance, and diameter in tree-structured graphical models.
CIKA uses LLM-based interventions to probe causal effects of concepts on math reasoning success, achieving competitive results on benchmarks like Omni-MATH and GSM8K with a frozen 7B model.
Derives annealed entropy rate, closed-form environment KL divergence, and quantitative bounds on trajectory KL convergence for edge-reinforced random walks using their random environment representation.
SLIM dynamically optimizes the active external skill set in agentic RL via leave-one-skill-out marginal contribution estimates and lifecycle operations, delivering a 7.1% average gain over baselines on ALFWorld and SearchQA while showing some skills remain externally useful.
Develops a McDiarmid-type concentration inequality for causal autoregressive processes that preserves sparsity to achieve O(1) variance proxies instead of O(N).
citing papers explorer
-
Population Risk Bounds for Kolmogorov-Arnold Networks Trained by DP-SGD with Correlated Noise
First population risk bounds for KANs under mini-batch DP-SGD with correlated noise, using a new non-convex optimization analysis combined with stability-based generalization.
-
Testing properties of trees in graphical models with covariance queries
The paper presents randomized tests with explicit query bounds for properties including number of leaves, maximum degree, typical distance, and diameter in tree-structured graphical models.
-
Mathematical Reasoning via Intervention-Based Time-Series Causal Discovery Using LLMs as Concept Mastery Simulators
CIKA uses LLM-based interventions to probe causal effects of concepts on math reasoning success, achieving competitive results on benchmarks like Omni-MATH and GSM8K with a frozen 7B model.
-
An Information-theoretic Analysis of Edge-reinforced Random Walks
Derives annealed entropy rate, closed-form environment KL divergence, and quantitative bounds on trajectory KL convergence for edge-reinforced random walks using their random environment representation.
-
Dynamic Skill Lifecycle Management for Agentic Reinforcement Learning
SLIM dynamically optimizes the active external skill set in agentic RL via leave-one-skill-out marginal contribution estimates and lifecycle operations, delivering a 7.1% average gain over baselines on ALFWorld and SearchQA while showing some skills remain externally useful.
-
Matrix-Decoupled Concentration for Autoregressive Sequences: Dimension-Free Guarantees for Sparse Long-Context Rewards
Develops a McDiarmid-type concentration inequality for causal autoregressive processes that preserves sparsity to achieve O(1) variance proxies instead of O(N).