analiza aiqbench
This research introduces a discrete denoising diffusion framework that utilizes the Neural Interaction Screening Estimator to model single-site conditional probabilities for generative tasks. By testing on synthetic Ising models and data from D-Wave quantum annealers, the study demonstrates superior performance in metrics like total variation compared to existing ratio-based methods. The work bridges advanced machine learning techniques with discrete state spaces relevant to both classical AI and quantum system simulation.
Discrete Diffusion with Sample-Efficient Estimators for Conditionals
arXiv:2602.20293v3 Announce Type: replace Abstract: We study a discrete denoising diffusion framework that integrates a sample-efficient estimator of single-site conditionals with round-robin noising and denoising dynamics for generative modeling over discrete state spaces. Rather than approximating a discrete analog of a score function, our formulation treats single-site conditional probabilities as the fundamental objects that parameterize the reverse diffusion process. We employ a sample-efficient method known as Neural Interaction Screening Estimator (NeurISE) to estimate these conditionals in the diffusion dynamics. Controlled experiments on synthetic Ising models, MNIST, and scientific data sets produced by a D-Wave quantum annealer, synthetic Potts model and one dimensional quantum systems demonstrate the proposed approach. On the binary data sets, these experiments demonstrate that the proposed approach outperforms popular existing methods including ratio-based approaches, achieving improved performance in total variation, cross-correlations, and kernel density estimation metrics.
Abstract:We study a discrete denoising diffusion framework that integrates a sample-efficient estimator of single-site conditionals with round-robin noising and denoising dynamics for generative modeling over discrete state spaces. Rather than approximating a discrete analog of a score function, our formulation treats single-site conditional probabilities as the fundamental objects that parameterize the reverse diffusion process. We employ a sample-efficient method known as Neural Interaction Screening Estimator (NeurISE) to estimate these conditionals in the diffusion dynamics. Controlled experiments on synthetic Ising models, MNIST, and scientific data sets produced by a D-Wave quantum annealer, synthetic Potts model and one dimensional quantum systems demonstrate the proposed approach. On the binary data sets, these experiments demonstrate that the proposed approach outperforms popular existing methods including ratio-based approaches, achieving improved performance in total variation, cross-correlations, and kernel density estimation metrics.
Machine Learning (cs.LG); Machine Learning (stat.ML)
arXiv:2602.20293 [cs.LG]
(or arXiv:2602.20293v3 [cs.LG] for this version)
Text integral sub licența deschisă a publicației · 148 cuvinte
Subiecte conexe
Fixed-point neural samplers on discrete spaces
arXiv · 2 oct. 2026
The Machines that Make the Machines
NVIDIA · 7 oct. 2026
Measurement incompatibility in Bayesian multiparameter quantum estimation
Quantum Journal · 13 aug. 2026
Quantum algorithms for general nonlinear dynamics based on the Carleman embedding
Quantum Journal · 8 oct. 2026