Total Variation Distance

Total variation distance (TVD) measures the difference between two probability distributions, quantifying how dissimilar they are. Current research focuses on efficiently estimating TVD in various contexts, including auditing the privacy guarantees of machine learning models, assessing the fidelity of generative data, and analyzing the convergence rates of sampling algorithms. These advancements are crucial for ensuring privacy in federated learning, improving the reliability of synthetic data, and developing more robust and efficient statistical methods across diverse fields. The ability to accurately and efficiently compute TVD has significant implications for both theoretical understanding and practical applications in machine learning, statistics, and beyond.

Papers

November 13, 2024

Bayesian Comparisons Between Representations
Heiko H. Schütt
Meaningful Representation Predictive Distribution Jensen Shannon Divergence Total Variation Distance Kernel Target Alignment Similarity Assessment

October 28, 2024

Robust Estimation for Kernel Exponential Families with Smoothed Total Variation Distances
Takafumi Kanamori, Kodai Yokoyama, Takayuki Kawashima
Robust Estimation Statistical Model Statistical Inference Robust Statistic Statistical Framework Total Variation Distance Smoothed Analysis

July 5, 2024

A Unified Learn-to-Distort-Data Framework for Privacy-Utility Trade-off in Trustworthy Federated Learning
Xiaojin Zhang, Mingcong Xu, Wei Chen
Federated Learning Privacy Preserving Privacy Utility Total Variation Distance Private Bayesian

June 7, 2024

Black Box Differential Privacy Auditing Using Total Variation Distance
Antti Koskela, Jafar Mohammadi
Differential Privacy Black Box Differential Privacy Guarantee Privacy Model Total Variation Distance ONE Shot

May 24, 2024

Discriminative Estimation of Total Variation Distance: A Fidelity Auditor for Generative Data
Lan Tao, Shirong Xu, Chi-Hua Wang, Namjoon Suh, Guang Cheng
Generative AI Fidelity Reward Discriminative Learning Synthetic Image Data Total Variation Distance Generative Fidelity

September 17, 2023

Total Variation Distance Estimation Is as Easy as Probabilistic Inference
Arnab Bhattacharyya, Sutanu Gayen, Kuldeep S. Meel, Dimitrios Myrisiotis, A. Pavan, N. V. Vinodchandran
Graphical Model Distance Estimation Total Variation Distance Randomized Approximation Probabilistic Inference Algorithm

April 13, 2023

A Polynomial Time, Pure Differentially Private Estimator for Binary Product Distributions
Vikrant Singhal
Sample Complexity Polynomial Time Product Distribution Efficient Algorithm Total Variation Distance

February 26, 2023

Tailoring Language Generation Models under Total Variation Distance
Haozhe Ji, Pei Ke, Zhipeng Hu, Rongsheng Zhang, Minlie Huang
Language Generation Natural Language Generation Text Generation Task Language Generation Model Autoregressive Decoding Total Variation Distance

February 23, 2023

Beyond Moments: Robustly Learning Affine Transformations with Asymptotically Optimal Error
He Jia, Pravesh K . Kothari, Santosh S. Vempala
Independent Component Analysis Temporal Moment Affine Transformation Optimal Error Total Variation Distance

February 9, 2023

The Sample Complexity of Approximate Rejection Sampling with Applications to Smoothed Online Learning
Adam Block, Yury Polyanskiy
Financial Application Sample Complexity Rejection Sampling F$ Divergence Total Variation Distance Independent Sampling Smoothed Online Learning

December 12, 2022

Lower Bounds for the Total Variation Distance Given Means and Variances of Distributions
Tomohiro Nishiyama
Lower Bound Multiple Meaning Covariance Matrix Probability Measure Total Variation Distance Data Variance

October 13, 2022

Condition-number-independent convergence rate of Riemannian Hamiltonian Monte Carlo with numerical integrators
Yunbum Kook, Yin Tat Lee, Ruoqi Shen, Santosh S. Vempala
Convergence Rate Hamiltonian Monte Carlo Total Variation Distance Numerical Integrator Constrained Distribution

September 15, 2022

The Impact of Edge Displacement Vaserstein Distance on UD Parsing Performance
Mark Anderson, Carlos Gómez-Rodríguez
Global Impact System Performance NLP Field Dependency Treebanks Total Variation Distance Statistical Correlation Balanced Split

March 22, 2022

Distributionally Robust Model Predictive Control with Total Variation Distance
Anushri Dixit, Mohamadreza Ahmadi, Joel W. Burdick
Predictive Control Model Predictive Control Distributionally Robust Total Variation Distance Chance Constrained Ambiguity Set

November 7, 2021

Sampling from Log-Concave Distributions with Infinity-Distance Guarantees
Oren Mangoubi, Nisheeth K. Vishnoi
Open Sampling Log Concave Total Variation Distance Variational Bound

Total Variation Distance

Papers

Bayesian Comparisons Between Representations

Robust Estimation for Kernel Exponential Families with Smoothed Total Variation Distances

A Unified Learn-to-Distort-Data Framework for Privacy-Utility Trade-off in Trustworthy Federated Learning

Black Box Differential Privacy Auditing Using Total Variation Distance

Discriminative Estimation of Total Variation Distance: A Fidelity Auditor for Generative Data

Total Variation Distance Estimation Is as Easy as Probabilistic Inference

A Polynomial Time, Pure Differentially Private Estimator for Binary Product Distributions

Tailoring Language Generation Models under Total Variation Distance

Beyond Moments: Robustly Learning Affine Transformations with Asymptotically Optimal Error

The Sample Complexity of Approximate Rejection Sampling with Applications to Smoothed Online Learning

Lower Bounds for the Total Variation Distance Given Means and Variances of Distributions

Condition-number-independent convergence rate of Riemannian Hamiltonian Monte Carlo with numerical integrators

The Impact of Edge Displacement Vaserstein Distance on UD Parsing Performance

Distributionally Robust Model Predictive Control with Total Variation Distance

Sampling from Log-Concave Distributions with Infinity-Distance Guarantees