A hybrid evolution-strategy and gradient-descent framework maximizes a non-differentiable 'surprise score' to discover non-random features for non-parametric self-supervised image clustering.
An intuitive proof of the data processing inequality
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
abstract
The data processing inequality (DPI) is a fundamental feature of information theory. Informally it states that you cannot increase the information content of a quantum system by acting on it with a local physical operation. When the smooth min-entropy is used as the relevant information measure, then the DPI follows immediately from the definition of the entropy. The DPI for the von Neumann entropy is then obtained by specializing the DPI for the smooth min-entropy by using the quantum asymptotic equipartition property (QAEP). We provide a new, simplified proof of the QAEP and therefore obtain a self-contained proof of the DPI for the von Neumann entropy.
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
SynGR is a synergistic generative recommendation method that masks dominant-modal tokens and uses contrastive learning to capture cross-modal synergies, outperforming prior methods on Amazon Arts, Games, and Instruments.
The Twisted-Path Particle Filter parameterizes twisting functions via neural networks and optimizes them against a path-measure KL divergence to improve continuous-time particle filtering.
InfoGeo reformulates cross-view geo-localization as an information bottleneck that aligns object-centric structural relations across views while suppressing view-specific noise.
A self-supervised approach uses consistent spatial relationships of anatomical structures across patients to improve 3D multi-modal medical image representations, yielding modest gains on segmentation and classification tasks.
TaTok is a theoretically grounded adaptive tokenization method that uses global tokens and cumulative conditional entropy filtering to reduce redundancy while improving reconstruction quality over fixed-rate patch tokenization.
citing papers explorer
-
Converge to Surprise: Evolutionary Self-supervised Image Clustering
A hybrid evolution-strategy and gradient-descent framework maximizes a non-differentiable 'surprise score' to discover non-random features for non-parametric self-supervised image clustering.
-
SynGR: Unleashing the Potential of Cross-Modal Synergy for Generative Recommendation
SynGR is a synergistic generative recommendation method that masks dominant-modal tokens and uses contrastive learning to capture cross-modal synergies, outperforming prior methods on Amazon Arts, Games, and Instruments.
-
Guidance for twisted particle filter: a continuous-time perspective
The Twisted-Path Particle Filter parameterizes twisting functions via neural networks and optimizes them against a path-measure KL divergence to improve continuous-time particle filtering.
-
InfoGeo: Information-Theoretic Object-Centric Learning for Cross-View Generalizable UAV Geo-Localization
InfoGeo reformulates cross-view geo-localization as an information bottleneck that aligns object-centric structural relations across views while suppressing view-specific noise.
-
Beyond Instance-Level Self-Supervision in 3D Multi-Modal Medical Imaging
A self-supervised approach uses consistent spatial relationships of anatomical structures across patients to improve 3D multi-modal medical image representations, yielding modest gains on segmentation and classification tasks.
-
Mutual Enhancement Between Global Tokens and Patch Tokens: From Theory to Practice
TaTok is a theoretically grounded adaptive tokenization method that uses global tokens and cumulative conditional entropy filtering to reduce redundancy while improving reconstruction quality over fixed-rate patch tokenization.