Convergence properties of Markov models for image generation with applications to spin-flip dynamics and to diffusion processes

Abstract

In the field of Markov models for image generation, the main idea is to learn how non-trivial images are gradually destroyed by a trivial forward Markov dynamics over the large time window $[0,t]$ converging towards pure noise for $t\to +\infty$, and to implement the non-trivial backward time-dependent Markov dynamics over the same time window $[0,t]$ starting from pure noise at $t$ in order to generate new images at time $0$. The goal of the present paper is to analyze the convergence properties of this reconstructive backward dynamics as a function of the time $t$ using the spectral properties of the trivial continuous-time forward dynamics for the $N$ pixels $n=1,..,N$. The general framework is applied to two cases : (i) when each pixel $n$ has only two states $S_n=\pm 1$ with Markov jumps between them; (ii) when each pixel $n$ is characterized by a continuous variable $x_n$ that diffuses on an interval $]x_{-},x_{+}[$ that can be either finite or infinite.

v2= revised version with new sections on illustrative examples (28 pages)

References in corpus (13)

• Quantum mechanical and information theoretic view on classical glass transitions
• Dynamical Regimes of Diffusion Models
• Localization of thermal packets and metastable states in Sinai model
• Eigenvalue method to compute the largest relaxation time of disordered systems
• Generative diffusion in very large dimensions
• Large deviations for the Skew-Detailed-Balance Lifted-Markov processes to sample the equilibrium distribution of the Curie-Weiss model
• One-dimensional classical diffusion in a random force field with weakly concentrated absorbers
• Optimal Diffusion Processes
• Large deviations for the Pearson family of ergodic diffusion processes involving a quadratic diffusion coefficient and a linear force
• Revisiting boundary-driven non-equilibrium Markov dynamics in arbitrary potentials via supersymmetric quantum mechanics and explicit large deviations at various levels
• On the Kemeny time for continuous-time reversible and irreversible Markov processes with applications to stochastic resetting and to conditioning towards forever-survival
• A supersymmetric quantum perspective on the explicit large deviations for reversible Markov jump processes, with applications to pure and random spin chains
• Markov spin models for image generation : explicit large deviations with respect to the number of pixels