DeepConvContext, a two-stage LSTM architecture, captures both intra-window and inter-window temporal patterns in inertial sensor data and reports roughly 10-point F1 improvement over DeepConvLSTM across six HAR benchmarks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
DeepConvContext: A Multi-Scale Approach to Timeseries Classification in Human Activity Recognition
DeepConvContext, a two-stage LSTM architecture, captures both intra-window and inter-window temporal patterns in inertial sensor data and reports roughly 10-point F1 improvement over DeepConvLSTM across six HAR benchmarks.