Timezone: »
Mini-batch optimal transport (m-OT) has been successfully used in practical applications that involve probability measures with a very high number of supports. The m-OT solves several smaller optimal transport problems and then returns the average of their costs and transportation plans. Despite its scalability advantage, the m-OT does not consider the relationship between mini-batches which leads to undesirable estimation. Moreover, the m-OT does not approximate a proper metric between probability measures since the identity property is not satisfied. To address these problems, we propose a novel mini-batch scheme for optimal transport, named Batch of Mini-batches Optimal Transport (BoMb-OT), that finds the optimal coupling between mini-batches and it can be seen as an approximation to a well-defined distance on the space of probability measures. Furthermore, we show that the m-OT is a limit of the entropic regularized version of the BoMb-OT when the regularized parameter goes to infinity. Finally, we carry out experiments on various applications including deep generative models, deep domain adaptation, approximate Bayesian computation, color transfer, and gradient flow to show that the BoMb-OT can be widely applied and performs well in various applications.
Author Information
Khai Nguyen (University of Texas at Austin)
Dang Nguyen (VinAI)

AI Resident
Quoc Nguyen (VinAI)
Tung Pham (VinAI Research)
Hung Bui (VinAI Research)
Dinh Phung (Monash University, Australia)
Trung Le (Monash University)
Nhat Ho (University of Texas at Austin)
Related Events (a corresponding poster, oral, or spotlight)
-
2022 Spotlight: On Transportation of Mini-batches: A Hierarchical Approach »
Wed. Jul 20th 02:45 -- 02:50 PM Room Ballroom 3 & 4
More from the Same Authors
-
2023 Poster: Vector Quantized Wasserstein Auto-Encoder »
Tung-Long Vuong · Trung Le · He Zhao · Chuanxia Zheng · Mehrtash Harandi · Jianfei Cai · Dinh Phung -
2023 Poster: Neural Collapse in Deep Linear Networks: From Balanced to Imbalanced Data »
Hien Dang · Tho Tran Huu · TAN NGUYEN · Stanley Osher · Hung Tran-The · Nhat Ho -
2023 Poster: Revisiting Over-smoothing and Over-squashing Using Ollivier-Ricci Curvature »
Khang Nguyen · Nong Hieu · Vinh NGUYEN · Nhat Ho · Stanley Osher · TAN NGUYEN -
2023 Poster: On Excess Mass Behavior in Gaussian Mixture Models with Orlicz-Wasserstein Distances »
Aritra Guha · Nhat Ho · XuanLong Nguyen -
2023 Poster: Self-Attention Amortized Distributional Projection Optimization for Sliced Wasserstein Point-Cloud Reconstruction »
Khai Nguyen · Dang Nguyen · Nhat Ho -
2022 Poster: Entropic Gromov-Wasserstein between Gaussian Distributions »
Khang Le · Dung Le · Huy Nguyen · · Tung Pham · Nhat Ho -
2022 Poster: Improving Transformers with Probabilistic Attention Keys »
Tam Nguyen · Tan Nguyen · Dung Le · Duy Khuong Nguyen · Viet-Anh Tran · Richard Baraniuk · Nhat Ho · Stanley Osher -
2022 Spotlight: Improving Transformers with Probabilistic Attention Keys »
Tam Nguyen · Tan Nguyen · Dung Le · Duy Khuong Nguyen · Viet-Anh Tran · Richard Baraniuk · Nhat Ho · Stanley Osher -
2022 Spotlight: Entropic Gromov-Wasserstein between Gaussian Distributions »
Khang Le · Dung Le · Huy Nguyen · · Tung Pham · Nhat Ho -
2022 Poster: Architecture Agnostic Federated Learning for Neural Networks »
Disha Makhija · Xing Han · Nhat Ho · Joydeep Ghosh -
2022 Poster: Improving Mini-batch Optimal Transport via Partial Transportation »
Khai Nguyen · Dang Nguyen · The-Anh Vu-Le · Tung Pham · Nhat Ho -
2022 Spotlight: Architecture Agnostic Federated Learning for Neural Networks »
Disha Makhija · Xing Han · Nhat Ho · Joydeep Ghosh -
2022 Spotlight: Improving Mini-batch Optimal Transport via Partial Transportation »
Khai Nguyen · Dang Nguyen · The-Anh Vu-Le · Tung Pham · Nhat Ho -
2022 Poster: Refined Convergence Rates for Maximum Likelihood Estimation under Finite Mixture Models »
Tudor Manole · Nhat Ho -
2022 Oral: Refined Convergence Rates for Maximum Likelihood Estimation under Finite Mixture Models »
Tudor Manole · Nhat Ho -
2021 Poster: Temporal Predictive Coding For Model-Based Planning In Latent Space »
Tung Nguyen · Rui Shu · Tuan Pham · Hung Bui · Stefano Ermon -
2021 Spotlight: Temporal Predictive Coding For Model-Based Planning In Latent Space »
Tung Nguyen · Rui Shu · Tuan Pham · Hung Bui · Stefano Ermon -
2021 Poster: LAMDA: Label Matching Deep Domain Adaptation »
Trung Le · Tuan Nguyen · Nhat Ho · Hung Bui · Dinh Phung -
2021 Spotlight: LAMDA: Label Matching Deep Domain Adaptation »
Trung Le · Tuan Nguyen · Nhat Ho · Hung Bui · Dinh Phung -
2020 Poster: Predictive Coding for Locally-Linear Control »
Rui Shu · Tung Nguyen · Yinlam Chow · Tuan Pham · Khoat Than · Mohammad Ghavamzadeh · Stefano Ermon · Hung Bui -
2020 Poster: On Unbalanced Optimal Transport: An Analysis of Sinkhorn Algorithm »
Khiem Pham · Khang Le · Nhat Ho · Tung Pham · Hung Bui -
2020 Poster: Parameterized Rate-Distortion Stochastic Encoder »
Quan Hoang · Trung Le · Dinh Phung