Probabilistic Inference and Differential Privacy

By Oliver Williams, Frank McSherry. 2010

146 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Fair decision making using privacy-protected data (2019, background): “Estimating posterior distributions from differentially-private statistics has been studied previously [9, 10, 33].”
Building a RAPPOR with the Unknown: Privacy-Preserving Learning of Associations and Data Dictionaries (2015, positive): “The question of estimating distributions from differentially private data is not new, with Williams et al. ﬁrst making explicit the connection between probabilistic inference and differential privacy [24, 13]. This previous work is similar in principle to our approach. There has even been some work ”
Encode, Shuffle, Analyze Privacy Revisited: Formalizations and Empirical Evaluation (2020, background): “At a high level, LDP-SGD follows the following template of noisy stochastic gradient descent [38], [39], [40].”
Privacy for Free: Posterior Sampling and Stochastic Gradient Monte Carlo (2015, background): “Integrating Williams & McSherry (2010) with our procedure might lead to some further performance boost, but investigating its eﬀect is beyond the scope of the current paper.”
A workflow for differentially-private graph synthesis (2012, methodology): “Following [14], these probabilities can lead us to a most likely dataset, or allow us to define a posterior distribution over datasets given some prior.”
Monitoring web browsing behavior with differential privacy (2014, methodology): “With the technique of posterior estimation, also referred to as probabilistic inference in [26], we have proposed FAST framework in [14, 15] for releasing time series data with user-level differential privacy.”
Near-Optimal Algorithms for Differentially-Private Principal Components (2012, background): “…motivated definition of privacy (Dwork et al., 2006b) that has gained significant attention over the past few years in the machine-learning and data-mining communities (Machanavajjhala et al., 2008; McSherry and Mironov, 2009; McSherry, 2009; Friedman and Schuster, 2010; Mohammed et al., 2011).”
Finite Sample Differentially Private Confidence Intervals (2017, background): “…the combined uncertainty due to sampling and privacy (for various differentially private algorithms) were given in Vu and Slavkovic (2009); Williams and McSherry (2010); Karwa and Slavković (2012); Karwa et al. (2015, 2014); Karwa and Slavković (2016); Gaboardi et al. (2016b); Solea…”
Is Interaction Necessary for Distributed Private Learning? (2017, methodology): “Noisy variants of firstorder methods play an important role in several of those works, starting with Williams and McSherry [47].”
Differentially Private Online Learning (2011, unknown): “In particular, there exists many results concerning privacy for learning problems [2, 3, 35, 29, 33].”
Stochastic gradient descent with differentially private updates (2013, methodology): “The PINQ [15] package uses a noisy sum operation to compute full noisy gradient steps for logistic regression [16].”
Differentially Private Distributed Online Learning (2015, background): “Many privacy-preserving algorithms [7, 8, 9] have been proposed to use differential privacy to protect sensitive information in the centralized ofﬂine learning framework.”
ML Confidential: Machine Learning on Encrypted Data (2012, background): “Note that the Confidential ML problem discussed in this paper is also loosely related to doing inference on differentially private data (see [21] and references therein), the difference being that in our case the Cloud Service performing the inference calculations is not even able to interpret the r”
Differentially Private High-Dimensional Data Publication via Sampling-Based Inference (2015, background): “The connection between probabilistic inference and differential privacy is also studied in [23].”
Differentially Private Bayesian Linear Regression (2019, methodology): “This motivates our development of Bayesian inference methods for linear regression that properly account for the noise due to the privacy mechanism [Bernstein and Sheldon, 2018, Bernstein et al., 2017, Karwa et al., 2014, 2016, Schein et al., 2018, Williams and McSherry, 2010].”