The Finite-Time Spectral Sensitivity g(t), a gradient-free, forward-pass metric that exposes flow geometry by tracking the root-mean-square singular value of the state-transition matrix, is presented.
Abstract
Continuous-time generative frameworks construct probability paths between base and target domains by optimizing time-dependent velocity fields. While theoretical targets favor straight trajectories, empirical networks develop complex path deformations. This paper presents the Finite-Time Spectral Sensitivity (FTSS) g(t), a gradient-free, forward-pass metric that exposes flow geometry by tracking the root-mean-square singular value of the state-transition matrix. Serving as a continuous proxy for stable rank, g(t) reveals a distinct geometric pathology under data scarcity: while generalizing models maintain stable effective dimensions, overfitting causes a spectral collapse. We leverage this structural phenomenon to develop an internal geometric audit based on g(t). Our framework detects generative memorization using purely internal trajectory dynamics, removing the need for external membership queries or baseline data comparison.
A data-driven framework for approximating the first same-direction return time and the associated Poincaré map of chaotic flows is developed, and a symbolic residual-learning strategy based on finite itineraries of transverse section crossings is introduced.
Meng-Zhao Jin, Jana Rodriguez Hertz, Hong-Kun Zhang· Chaos· 0 citations
Intrinsic explainability remains a challenging problem, particularly in contexts where multilayer perceptrons (MLPs) require dynamic re-training within an optimization environment. This paper investigates how MLPs and their training dynamics can be represented and studied in non-Euclidean spaces; our representation fea...
Two lightweight, training-free algorithms are proposed, CAT-OV and CAT-OT that adapt step-sizes at inference time based on a novel connection between Flow Matching sampling and gradient flow that outperform existing step-size heuristics in image quality metrics across four text- to-image Flow Matching models.
Qinchan Li, Pedro Cisneros-Velarde, Ke-Ru Fu et al.· 0 citations
State space models (SSMs) have emerged as a compelling alternative to transformers for visual recognition, offering linear computational complexity while maintaining competitive accuracy. However, the lack of interpretability tools designed specifically for the recurrent dynamics of these models remains a significant g...
M. Mabrok, Yalda Zafari· IEEE Transactions on Neural...· 2 citations
GeometricSPRINT (Geometric Step Pruning for Inference in Trajectories), a training-free framework for constructing non-uniform sampling schedules from the geometry of denoising trajectories, consistently improves over uniform DDIM (Denoising Diffusion Implicit Models) schedules at matched NFE budgets.
In nonlinear prediction, two state representations with the same local uncertainty volume can have radically different predictive value. We consider differentiable finite-time dynamics together with a quadratic terminal requirement that specifies which terminal state differences matter. Pulling this requirement back th...
Song-Ju Kim· 2 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
