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

We describe a novel metric-based learning approach that introduces amultimodal framework and uses deep audio and geophone encoders in siameseconfiguration to design an adaptable and lightweight supervised model. Thisframework eliminates the need for expensive data labeling procedures and learnsgeneral-purpose representations from low multisensory data obtained fromomnipresent sensing systems. These sensing systems provide numerousapplications and various use cases in activity recognition tasks. Here, weintend to explore the human footstep movements from indoor environments andanalyze representations from a small self-collected dataset of acoustic andvibration-based sensors. The core idea is to learn plausible similaritiesbetween two sensory traits and combining representations from audio andgeophone signals. We present a generalized framework to learn embeddings fromtemporal and spatial features extracted from audio and geophone signals. Wethen extract the representations in a shared space to maximize the learning ofa compatibility function between acoustic and geophone features. This, in turn,can be used effectively to carry out a classification task from the learnedmodel, as demonstrated by assigning high similarity to the pairs with a humanfootstep movement and lower similarity to pairs containing no footstepmovement. Performance analyses show that our proposed multimodal frameworkachieves a 19.99\% accuracy increase (in absolute terms) and avoidedoverfitting on the evaluation set when the training samples were increased from200 pairs to just 500 pairs while satisfactorily learning the audio andgeophone representations. Our results employ a metric-based contrastivelearning approach for multi-sensor data to mitigate the impact of data scarcityand perform human movement identification with limited data size.

Metric-based multimodal meta-learning for human movement identification via footstep recognition