Diffusion Embeddings

Diffusion embeddings are techniques for representing high-dimensional data in lower dimensions by leveraging diffusion processes, often inspired by concepts from graph theory and optimal transport. Current research focuses on developing novel algorithms, such as those incorporating reaction-diffusion processes or optimal transport, to improve the accuracy and robustness of these embeddings, particularly for handling multiple datasets or time series data. These methods find applications in diverse fields, including data visualization, recommendation systems, and single-cell analysis, by revealing underlying data structures and facilitating tasks like data integration, anomaly detection, and clustering. The ability to learn continuous manifold representations from discrete data, as explored in recent work, further enhances the utility and interpretability of diffusion embeddings.

Papers

October 18, 2024

Mitigating Embedding Collapse in Diffusion Models for Categorical Data
Bac Nguyen, and Chieh-Hsin Lai, Yuhta Takida, Naoki Murata, Toshimitsu Uesaka, Stefano Ermon, Yuki Mitsufuji
Diffusion Model Latent Diffusion Model Categorical Data Event Collapse Diffusion Embeddings

July 1, 2024

Entropic Optimal Transport Eigenmaps for Nonlinear Alignment and Joint Embedding of High-Dimensional Datasets
Boris Landa, Yuval Kluger, Rong Ma
High Dimensional High Dimensional Data Low Dimensional Manifold Joint Embedding Entropic Optimal Transport Laplacian Eigenmaps Aware Alignment Diffusion Embeddings

May 9, 2024

Self-Supervised Learning of Time Series Representation via Diffusion Process and Imputation-Interpolation-Forecasting Mask
Zineb Senane, Lele Cao, Valentin Leonhard Buchner, Yusuke Tashiro, Lei You, Pawel Herman, Mats Nordahl, Ruibo Tu, Vilhelm von Ehrenheim
Diffusion Model Self Supervised Learning Mask Frozen DETR Diffusion Process Diffusion Based Method Time Series Representation Learning Time Series Representation Diffusion Embeddings

December 27, 2023

RDGCL: Reaction-Diffusion Graph Contrastive Learning for Recommendation
Jeongwhan Choi, Hyowon Wi, Chaejeong Lee, Sung-Bae Cho, Dongha Lee, Noseong Park
Contrastive Learning Recommender System Effective Recommendation Graph Contrastive Learning GNN Based Diffusion Embeddings

June 1, 2023

Neural FIM for learning Fisher Information Metrics from point cloud data
Oluwadamilola Fasina, Guillaume Huguet, Alexander Tong, Yanlei Zhang, Guy Wolf, Maximilian Nickel, Ian Adelstein, Smita Krishnaswamy
Statistical Manifold Cluster Center Fisher Information Metric Diffusion Embeddings

Diffusion Embeddings

Papers

Mitigating Embedding Collapse in Diffusion Models for Categorical Data

Entropic Optimal Transport Eigenmaps for Nonlinear Alignment and Joint Embedding of High-Dimensional Datasets

Self-Supervised Learning of Time Series Representation via Diffusion Process and Imputation-Interpolation-Forecasting Mask

RDGCL: Reaction-Diffusion Graph Contrastive Learning for Recommendation

Neural FIM for learning Fisher Information Metrics from point cloud data