Symmetric Norm Estimation and Regression on Sliding Windows

@article{Braverman2021SymmetricNE,
  title={Symmetric Norm Estimation and Regression on Sliding Windows},
  author={Vladimir Braverman and Viska Wei and Samson Zhou},
  journal={ArXiv},
  year={2021},
  volume={abs/2109.01635}
}
The sliding window model generalizes the standard streaming model and often performs better in applications where recent data is more important or more accurate than data that arrived prior to a certain time. We study the problem of approximating symmetric norms (a norm on R that is invariant under sign-flips and coordinate-wise permutations) in the sliding window model, where only the W most recent updates define the underlying frequency vector. Whereas standard norm estimation algorithms for… 

References

SHOWING 1-10 OF 62 REFERENCES
Near Optimal Linear Algebra in the Online and Sliding Window Models
TLDR
A unified row-sampling based framework that gives randomized algorithms for spectral approximation, low-rank approximation/projection-cost preservation, and $\ell_{1}$-subspace embeddings in the sliding window model, which often use nearly optimal space and achieve nearly input sparsity runtime.
Almost-Smooth Histograms and Sliding-Window Graph Algorithms
TLDR
The smooth-histogram framework of Braverman and Ostrovsky (FOCS 2007) is extended to almost-smooth functions, which includes all subadditive functions, and it is shown that if a sub additive function can be $(1+\epsilon)-approximated in the insertion-only streaming model, then it can be$.
Smooth Histograms for Sliding Windows
TLDR
This paper presents a new smooth histograms method that improves the approximation error rate obtained via exponential histograms and provides the first approximation algorithms for the following functions: Lp norms for p notin, frequency moments, length of increasing subsequence and geometric mean.
Clustering Problems on Sliding Windows
TLDR
A data structure that extends smooth histograms as introduced by Braverman and Ostrovsky to operate on a broader class of functions and shows that using only polylogarithmic space the authors can maintain a summary of the current window from which they can construct an O(1)-approximate clustering solution.
Better Sliding Window Algorithms to Maximize Subadditive and Diversity Objectives
TLDR
This work describes an alternative approach to designing efficient sliding window algorithms for maximization problems, and instantiates this approach on a wide range of problems, yielding better algorithms for submodular function optimization, diversity optimization and general subadditive optimization.
Optimal sampling from sliding windows
TLDR
This paper formally introduced the problem of sampling from sliding windows and improved the over-sampling method for sampling with replacement and provides optimal sampling schemas for all variants of the problem, i.e., sampling with or without replacement from fixed or bursty windows.
Separations for Estimating Large Frequency Moments on Data Streams
TLDR
If the updates to the vector arrive in the random-order insertion-only model, then there exist space efficient algorithms with improved dependencies on the approximation parameter ε, which gives an improved lower bound of Ω ( 1 ε · n ) for one-pass insertiononly streams.
Tight Bounds for Adversarially Robust Streams and Sliding Windows via Difference Estimators
TLDR
The results show there is no separation between the sliding window model and the standard data stream model in terms of the approximation factor, and the first difference estimators for a wide range of problems are developed.
Matrix Sketching Over Sliding Windows
TLDR
It is shown that maintaining ATA exactly requires linear space in the sliding window model, as opposed to O(d2) space inThe streaming model, and three general frameworks for matrix sketching on sliding windows are presented.
Truly Perfect Samplers for Data Streams and Sliding Windows
TLDR
This work shows that sublinear space truly perfect sampling is impossible in the turnstile model, and proves a lower bound of Ω(min(n, log 1/γ) for any G-sampler with point-wise error γ from the true distribution, and gives a general time-efficient sublinear-space framework for developing truly perfect samplers in the insertion-only streaming and sliding window models.
...
...