Decomposes pre-softmax attention QK^T into symmetric and skew-symmetric components to derive Hopfield stability measures that correlate with fidelity-diversity in diffusion generation and introduces a circulation-based modulation knob.
Neural networks and physical systems with emergent collective computational abilities
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.LG 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
Thesis uses statistical mechanics to study DAM and RBM models for understanding memorization, low-dimensional learning, and adversarial robustness in neural networks.
citing papers explorer
-
Balancing Fidelity and Diversity in Diffusion Models via Symmetric Attention Decomposition: Hopfield Perspective
Decomposes pre-softmax attention QK^T into symmetric and skew-symmetric components to derive Hopfield stability measures that correlate with fidelity-diversity in diffusion generation and introduces a circulation-based modulation knob.
-
Explaining Machine Learning and Memorization with Statistical Mechanics
Thesis uses statistical mechanics to study DAM and RBM models for understanding memorization, low-dimensional learning, and adversarial robustness in neural networks.