Part IV: Deep Learning for Sensor Time Series
Chapter 15  [C][A]

Transformers for Sensor Data

Placeholder chapter page. Sections are produced by the book-skills pipeline.

Sections

  1. 15.1 Attention over time and channels
  2. 15.2 Patchification of sensor streams
  3. 15.3 Positional encoding for time and sensor identity
  4. 15.4 Long-context and efficient attention
  5. 15.5 Channel-independent vs channel-mixing designs
  6. 15.6 Masked sensor modeling
  7. 15.7 Compute-efficient deployment

Lab 15

train a patch transformer for sensor classification; compare with CNN/TCN baselines.