Towards Differential Relational Privacy and its use in Question Answering
Not on the map. Why?
Abstract (arXiv)
Memorization of the relation between entities in a dataset can lead to privacy issues when using a trained model for question answering. We introduce Relational Memorization (RM) to understand, quantify and control this phenomenon. While bounding general memorization can have detrimental effects on the performance of a trained model, bounding RM does not prevent effective learning. The difference is most pronounced when the data distribution is long-tailed, with many queries having only few training examples: Impeding general memorization prevents effective learning, while impeding only relational memorization still allows learning general properties of the underlying concepts. We formalize the notion of Relational Privacy (RP) and, inspired by Differential Privacy (DP), we provide a possible definition of Differential Relational Privacy (DrP). These notions can be used to describe and compute bounds on the amount of RM in a trained model. We illustrate Relational Privacy concepts in experiments with large-scale models for Question Answering.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 23 of its 28 references that are in the map. First posted on arXiv 30 Mar 2022.
Builds on
23 of 28 references.
References · 23
- 2020Language Models are Few-Shot Learnerscited by 32,199
- 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,491
- 2016SQuAD: 100,000+ Questions for Machine Comprehension of Textcited by 4,115
- 2016Deep Learning with Differential Privacycited by 2,682
- 2018Neural Tangent Kernel: Convergence and Generalization in Neural Networkscited by 2,597
- 2016Membership Inference Attacks against Machine Learning Modelscited by 2,221
- 2020Extracting Training Data from Large Language Modelscited by 2,144
- 2019DROP: A Reading Comprehension Benchmark Requiring Discrete Reasoning Over Paragraphscited by 883
- 2018The Secret Sharer: Evaluating and Testing Unintended Memorization in Neural Networkscited by 847
- 2019Wide Neural Networks of Any Depth Evolve as Linear Models Under Gradient Descentcited by 732
- 2020What Neural Networks Memorize and Why: Discovering the Long Tail via Influence Estimationcited by 492
- 2019Does Learning Require Memorization? A Short Tale about a Long Tailcited by 452
11 more
- 2019White-box vs Black-box: Bayes Optimal Strategies for Membership Inferencecited by 265
- 2020Differentially Private Learning Needs Better Features (or Much More Data)cited by 233
- 2019Differentially Private Learning with Adaptive Clippingcited by 218
- 2021The Curious Case of Hallucinations in Neural Machine Translationcited by 161
- 2021Large Scale Private Learning via Low-rank Reparametrizationcited by 87
- 2021Selective Differential Privacy for Language Modelingcited by 61
- 2019NumNet: Machine Reading Comprehension with Numerical Reasoningcited by 59
- 2016Differential Privacy as a Mutual Information Constraintcited by 41
- 2020LQF: Linear Quadratic Fine-Tuningcited by 25
- 2020Reducing Quantity Hallucinations in Abstractive Summarizationcited by 22
- 2021Estimating informativeness of samples with Smooth Unique Informationcited by 15
Built on by
Citing papers per year: 2023: 1.
1 paper.