Part IV: Deep Learning for Sensor Time Series
Chapter 16  [A][R]

State-Space Models: Mamba, S4, and Linear-Time Sequences

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

Sections

  1. 16.1 Why quadratic attention hurts on long sensor sequences
  2. 16.2 Structured state-space models (S4, S5)
  3. 16.3 Selective state spaces (Mamba)
  4. 16.4 Time-series SSM variants (SiMBA, TimeMachine, Bi-Mamba)
  5. 16.5 SSMs vs transformers: accuracy, latency, memory, on-device fit
  6. 16.6 Hybrid SSM-attention architectures
  7. 16.7 When to reach for an SSM

Lab 16

benchmark a Mamba-style model vs a transformer on long-horizon sensor sequences (accuracy vs latency vs memory).