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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

In tasks such as tracking, time-series data inevitably carry missingobservations. While traditional tracking approaches can handle missingobservations, recurrent neural networks (RNNs) are designed to receive inputdata in every step. Furthermore, current solutions for RNNs, like omitting themissing data or data imputation, are not sufficient to account for theresulting increased uncertainty. Towards this end, this paper introduces anRNN-based approach that provides a full temporal filtering cycle for motionstate estimation. The Kalman filter inspired approach, enables to deal withmissing observations and outliers. For providing a full temporal filteringcycle, a basic RNN is extended to take observations and the associated beliefabout its accuracy into account for updating the current state. An RNNprediction model, which generates a parametrized distribution to capture thepredicted states, is combined with an RNN update model, which relies on theprediction model output and the current observation. By providing the modelwith masking information, binary-encoded missing events, the model can overcomelimitations of standard techniques for dealing with missing input values. Themodel abilities are demonstrated on synthetic data reflecting prototypicalpedestrian tracking scenarios.

Handling Missing Observations with an RNN-based Prediction-Update Cycle