AI Research
Researchers establish mathematical framework for understanding how generative AI models can survive training on contaminated data, offering crucial insights for maintaining synthetic media quality.
differential privacy
New research introduces risk-equalized differentially private synthetic data that protects outliers by controlling record-level influence, addressing critical privacy gaps in AI training data.
Synthetic Data
New research introduces PRISM, a differentially private synthetic data framework using structure-aware budget allocation to optimize prediction accuracy while maintaining privacy guarantees.
Synthetic Data
Researchers unveil new membership inference attack techniques for multi-table synthetic data, exposing privacy vulnerabilities in relational database anonymization systems.
AI Agents
New research presents a framework for building capable small language model agents using synthetic tasks, simulated environments, and structured rubric-based rewards—democratizing agentic AI development.
Synthetic Data
NVIDIA partners with Singapore to create culturally-aware synthetic training data using Nemotron personas, advancing sovereign AI development through co-designed data generation methodologies.
Synthetic Data
New survey introduces systematic metrics for evaluating synthetic data quality and trustworthiness from LLMs, addressing critical challenges in detecting and assessing AI-generated content reliability.
Synthetic Data
Researchers analyze why Empirical Risk Minimization fails when models train on synthetic data, revealing fundamental barriers that affect AI video generation and deepfake systems.
LLM unlearning
New research introduces domain-to-instance framework for generating synthetic data to help large language models selectively forget harmful knowledge while preserving useful capabilities.
Synthetic Data
New research explores how reinforcement learning can optimize synthetic data generation, with implications for training more capable AI video and media generation models.
Diffusion Models
New research reveals how diffusion models suffer 'generative collapse' when trained on synthetic data, with dominated samples disappearing while dominating ones proliferate across generations.
Diffusion Models
New research introduces SD2AIL, combining diffusion models with adversarial imitation learning to generate synthetic expert demonstrations, advancing AI training without human data dependency.