Differential privacy under continual observation

By Cynthia Dwork, Moni Naor, Toniann Pitassi, Guy N. Rothblum. 2010

845 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Privacy & Security

How researchers see this paper

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

In Artificial Intelligence

In Privacy & Security

Citations in the corpus

Practical and Private (Deep) Learning without Sampling or Shuffling (2021, methodology): “ing the now standard tree aggregation protocol (Dwork et al., 2010; Chan et al., 2011).”
The power of synergy in differential privacy: Combining a small curator with local randomizers (2019, neutral): “, [26].”
Federated Heavy Hitters Discovery with Differential Privacy (2019, background): “…some of these risks, the service provider can use the central model of diﬀerential privacy (DP) to provide internal or external analysts with a privacy-preserving set of learned heavy hitters (Dwork et al., 2006b,a; Dwork, 2008; Dwork et al., 2010; Bhaskar et al., 2010; Dwork and Roth, 2014).”
Amplification by Shuffling: From Local to Central Differential Privacy via Anonymity (2018, methodology): “Our algorithm is based on the tree-based aggregation scheme [DNPR10, CSS11] used previously for releasing continual statistics over longitudinal data in the central model.”
Collecting Telemetry Data Privately (2017, background): “nt line of work about streaming model of DP [14], the event-level private counting problem under continual observation is studied [11], with almost tight upper bounds in error (polynomial in logT) in [13] and [7]. [24] proposes a weaker protection for continual events in wconsecutive timestamps to r”
Local, Private, Efficient Protocols for Succinct Histograms (2015, background): “Running many copies of the protocol for unique heavy hitters in parallel, we can recover the list of heavy hitters.”
RAPPOR: Randomized Aggregatable Privacy-Preserving Ordinal Response (2014, background): “Although it does not operate under the same local model as RAPPOR, recent work on pan-private streaming and on privacy under continual observation introduces additional ideas relevant for the longitudinal data collection with privacy [13, 14].”
Differentially Private Distributed Optimization (2014, background): “Our definition of privacy is a modification of the notion of differential privacy introduced in [7] in the context of streaming algorithms.”
Knock Knock, Who's There? Membership Inference on Aggregate Location Data (2017, background): “This is not surprising, as previous work highlights the difficulty of releasing private statistics under continual observation [6, 10, 17].”
Advanced Probabilistic Couplings for Differential Privacy (2016, neutral): “The differential literature proposes several interactive algorithms; examples include private counters [14, 21], the Sparse Vector mechanism, and other algorithms using these mechanisms [30].”
Privacy via the Johnson-Lindenstrauss Transform (2012, background): “The work of [13] oﬀers directions for possibly overcoming this challenge.”
Proving Differential Privacy in Hoare Logic (2014, background): “On a more practical side, it would be interesting to implement our transformation for a realistic setting, for instance modeling thePINQ language [30].”
Differential Privacy Techniques for Cyber Physical Systems: A Survey (2018, background): “For instance, dealing with time-series data publishing, w-event privacy provides an optimal balance between event-level and user-level privacy [340].”
Monitoring web browsing behavior with differential privacy (2014, background): “Among these works two major research directions have been developed investigating the private release of numeric time series [23, 7, 11, 14, 22] and discrete series (event sequences) [16, 25] respectively.”
Is Interaction Necessary for Distributed Private Learning? (2017, methodology): “Our one-dimensional algorithm is based on a novel reduction of the general one-dimensional convex optimization to the median problem (a special case), followed by a noninteractive local algorithm for the median that uses a tree-based technique for simultaneously approximating 1D range queries (that ”