English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

The need to estimate a particular quantile of a distribution is an importantproblem which frequently arises in many computer vision and signal processingapplications. For example, our work was motivated by the requirements of manysemi-automatic surveillance analytics systems which detect abnormalities inclose-circuit television (CCTV) footage using statistical models of low-levelmotion features. In this paper we specifically address the problem ofestimating the running quantile of a data stream with non-stationarystochasticity when the memory for storing observations is limited. We makeseveral major contributions: (i) we derive an important theoretical resultwhich shows that the change in the quantile of a stream is constrainedregardless of the stochastic properties of data, (ii) we describe a set ofhigh-level design goals for an effective estimation algorithm that emerge as aconsequence of our theoretical findings, (iii) we introduce a novel algorithmwhich implements the aforementioned design goals by retaining a sample of datavalues in a manner adaptive to changes in the distribution of data andprogressively narrowing down its focus in the periods of quasi-stationarystochasticity, and (iv) we present a comprehensive evaluation of the proposedalgorithm and compare it with the existing methods in the literature on bothsynthetic data sets and three large `real-world' streams acquired in the courseof operation of an existing commercial surveillance system. Our findingsconvincingly demonstrate that the proposed method is highly successful andvastly outperforms the existing alternatives, especially when the targetquantile is high valued and the available buffer capacity severely limited.

The adaptable buffer algorithm for high quantile estimation in non-stationary data streams