Dywave uses wavelet hierarchical decomposition to create event-aligned compact token sequences for heterogeneous IoT signals, yielding up to 12% accuracy gains and 75% shorter inputs on mainstream sequence models across five datasets.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 2
citation-polarity summary
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1roles
background 1representative citing papers
citing papers explorer
-
Dywave: Event-Aligned Dynamic Tokenization for Heterogeneous IoT Sensing Signals
Dywave uses wavelet hierarchical decomposition to create event-aligned compact token sequences for heterogeneous IoT signals, yielding up to 12% accuracy gains and 75% shorter inputs on mainstream sequence models across five datasets.