DyWPE generates positional embeddings for time series transformers from the input signal via Discrete Wavelet Transform and outperforms standard positional encodings on ten datasets, especially longer sequences and biomedical signals.
arXiv preprint arXiv:2103.14438 (2021)
4 Pith papers cite this work. Polarity classification is still indexing.
verdicts
UNVERDICTED 4representative citing papers
GenHAR generalizes cross-domain human activity recognition by 9.97% accuracy and 6.4x lower FLOPs via tokenized sensor data, frequency channel correlations, selective masking, and efficient attention, with deployment detecting 2.15 billion activities.
Gated-CNN applies independent 1D convolutions and sigmoid gating to IMU streams from smartwatches, achieving 90-93% F1 on five datasets and 97% F1 with zero missed falls in real-time Pixel Watch testing, outperforming Transformer baselines.
A survey of positional encoding methods in transformer-based time series models that evaluates fixed, learnable, relative, and hybrid approaches on classification tasks and links effectiveness to data characteristics.
citing papers explorer
-
DyWPE: Signal-Aware Dynamic Wavelet Positional Encoding for Time Series Transformers
DyWPE generates positional embeddings for time series transformers from the input signal via Discrete Wavelet Transform and outperforms standard positional encodings on ten datasets, especially longer sequences and biomedical signals.
-
GenHAR: Generalizing Cross-domain Human Activity Recognition for Last-mile Delivery
GenHAR generalizes cross-domain human activity recognition by 9.97% accuracy and 6.4x lower FLOPs via tokenized sensor data, frequency channel correlations, selective masking, and efficient attention, with deployment detecting 2.15 billion activities.
-
You Don't Need Attention: Gated Convolutional Modeling for Watch-Based Fall Detection
Gated-CNN applies independent 1D convolutions and sigmoid gating to IMU streams from smartwatches, achieving 90-93% F1 on five datasets and 97% F1 with zero missed falls in real-time Pixel Watch testing, outperforming Transformer baselines.
-
Positional Encoding in Transformer-Based Time Series Models: A Survey
A survey of positional encoding methods in transformer-based time series models that evaluates fixed, learnable, relative, and hybrid approaches on classification tasks and links effectiveness to data characteristics.