Chen
A researcher who in 2026 proposed a Bernoulli factory method for converting clipped estimates of a Poisson distribution into per-token accept/reject decisions.
Timeline (last 90 days)
SIRS uses the Bernoulli factory proposed by Chen et al. (2026) to convert Poisson-distributed clipped estimates into per-token accept/reject decisions.
Research published by Kashani, Chen, and Ozcan proposes a hybrid deepfake detection classifier that transfers part of the inference process from the digital domain to a physical optical system.
Neural Ordinary Differential Equations (Neural ODE) were proposed by Chen et al. at NeurIPS 2018, parameterizing the continuous-time derivative of the system state with a neural network and integrating via an ODE solver.
A commentary article published on arXiv (arXiv:2609.15996) challenges Chen, Zhao, and Cohan's distributional evaluation of research ideas generated by large language models.
The strict verification rules of traditional speculative decoding were independently proposed by Leviathan et al. and Chen et al. in 2023, based on a rejection sampling verification protocol.
All Facts (5)
SIRS uses the Bernoulli factory proposed by Chen et al. (2026) to convert Poisson-distributed clipped estimates into per-token accept/reject decisions.
50%UnverifiedResearch published by Kashani, Chen, and Ozcan proposes a hybrid deepfake detection classifier that transfers part of the inference process from the digital domain to a physical optical system.
50%UnverifiedNeural Ordinary Differential Equations (Neural ODE) were proposed by Chen et al. at NeurIPS 2018, parameterizing the continuous-time derivative of the system state with a neural network and integrating via an ODE solver.
50%UnverifiedThe strict verification rules of traditional speculative decoding were independently proposed by Leviathan et al. and Chen et al. in 2023, based on a rejection sampling verification protocol.
50%UnverifiedA commentary article published on arXiv (arXiv:2609.15996) challenges Chen, Zhao, and Cohan's distributional evaluation of research ideas generated by large language models.
50%