Image de couverture de livre

Can Generative Artificial Intelligence Survive Data Contamination? Theoretical Guarantees under Contaminated Recursive Training

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Wang, Kevin, Niu, Hongqian, Li, Didong
Format:	Preprint
Publié:	2026
Sujets:	Machine Learning Artificial Intelligence Statistics Theory
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Understanding Overparametrization in Survival Models through Interpolation
par: Liu, Yin, et autres
Publié: (2025)

Minimax Statistical Estimation under Wasserstein Contamination
par: Chao, Patrick, et autres
Publié: (2023)

On the Optimality of the Median-of-Means Estimator under Adversarial Contamination
par: de Juan, Xabier, et autres
Publié: (2025)

On the Provable Performance Guarantee of Efficient Reasoning Models
par: Zeng, Hao, et autres
Publié: (2025)

Tail-Aware Information-Theoretic Generalization for RLHF and SGLD
par: Zhang, Huiming, et autres
Publié: (2026)

Neural Networks Learn Generic Multi-Index Models Near Information-Theoretic Limit
par: Zhang, Bohan, et autres
Publié: (2025)

A Computational Theory for Efficient Mini Agent Evaluation with Causal Guarantees
par: Yan, Hedong
Publié: (2025)

Training Implicit Generative Models via an Invariant Statistical Loss
par: de Frutos, José Manuel, et autres
Publié: (2024)

From Data-Driven to Purpose-Driven Artificial Intelligence: Systems Thinking for Data-Analytic Automation of Patient Care
par: Anadria, Daniel, et autres
Publié: (2025)

From Collapse to Improvement: Statistical Perspectives on the Evolutionary Dynamics of Iterative Training on Contaminated Sources
par: Bakshi, Soham, et autres
Publié: (2026)

Neural Networks Generalize on Low Complexity Data
par: Chatterjee, Sourav, et autres
Publié: (2024)

On the Statistical Properties of Generative Adversarial Models for Low Intrinsic Data Dimension
par: Chakraborty, Saptarshi, et autres
Publié: (2024)

Heavy-tailed Contamination is Easier than Adversarial Contamination
par: Cherapanamjeri, Yeshwanth, et autres
Publié: (2024)

O(d/T) Convergence Theory for Diffusion Probabilistic Models under Minimal Assumptions
par: Li, Gen, et autres
Publié: (2024)

Generalization Properties of Score-matching Diffusion Models for Intrinsically Low-dimensional Data
par: Chakraborty, Saptarshi, et autres
Publié: (2026)

When Can We Reuse a Calibration Set for Multiple Conformal Predictions?
par: Balinsky, A. A., et autres
Publié: (2025)

A Quantitative Characterization of Forgetting in Post-Training
par: Balasubramanian, Krishnakumar, et autres
Publié: (2026)

Search-Time Data Contamination
par: Han, Ziwen, et autres
Publié: (2025)

Cyclic Counterfactuals under Shift-Scale Interventions
par: Saha, Saptarshi, et autres
Publié: (2025)

High-Dimensional Gaussian Mean Estimation under Realizable Contamination
par: Diakonikolas, Ilias, et autres
Publié: (2026)

Generalising realisability in statistical learning theory under epistemic uncertainty
par: Cuzzolin, Fabio
Publié: (2024)

Guaranteed Recovery of Unambiguous Clusters
par: Mazooji, Kayvon, et autres
Publié: (2025)

Le Cam Distortion: A Decision-Theoretic Framework for Robust Transfer Learning
par: Akdemir, Deniz
Publié: (2025)

Towards Bayesian Data Selection
par: Rodemann, Julian
Publié: (2024)

Anomaly Detection with Adaptive and Aggressive Rejection for Contaminated Training Data
par: Lee, Jungi, et autres
Publié: (2025)

Robust Sparse Estimation for Gaussians with Optimal Error under Huber Contamination
par: Diakonikolas, Ilias, et autres
Publié: (2024)

Deep Generative Models: Complexity, Dimensionality, and Approximation
par: Wang, Kevin, et autres
Publié: (2025)

Compression, Generalization and Learning
par: Campi, Marco C., et autres
Publié: (2023)

Scaling Laws in Linear Regression: Compute, Parameters, and Data
par: Lin, Licong, et autres
Publié: (2024)

Convergence of Shallow ReLU Networks on Weakly Interacting Data
par: Dana, Léo, et autres
Publié: (2025)

On the Statistical Capacity of Deep Generative Models
par: Tam, Edric, et autres
Publié: (2025)

Maximizing the Potential of Synthetic Data: Insights from Random Matrix Theory
par: Firdoussi, Aymane El, et autres
Publié: (2024)

Generalization and Scaling Laws for Mixture-of-Experts Transformers
par: Mayaki, Mansour Zoubeirou a
Publié: (2026)

Counterfactual Generative Modeling with Variational Causal Inference
par: Wu, Yulun, et autres
Publié: (2024)

A Statistical Analysis of Deep Federated Learning for Intrinsically Low-dimensional Data
par: Chakraborty, Saptarshi, et autres
Publié: (2024)

Improved Guarantees for Heterogeneous Treatment-Effect Estimation via Matrix Completion
par: Mehrotra, Anay, et autres
Publié: (2026)

A Theory of the Mechanics of Information: Generalization Through Measurement of Uncertainty (Learning is Measuring)
par: Hazard, Christopher J., et autres
Publié: (2025)

A Score-Based Density Formula, with Applications in Diffusion Generative Models
par: Li, Gen, et autres
Publié: (2024)

Labels or Preferences? Budget-Constrained Learning with Human Judgments over AI-Generated Outputs
par: Dong, Zihan, et autres
Publié: (2026)

U-Nets as Belief Propagation: Efficient Classification, Denoising, and Diffusion in Generative Hierarchical Models
par: Mei, Song
Publié: (2024)