VS
Energy
Sparsity • firing rate
Compare models
- LSTMSequential baseline
- TCNConvolutional backbone
- SNN‑onlyEvent‑driven
- Hybrid TCN–SNNParallel fusion
- SpikingTCNSpikes in conv blocks
Encoding methods
Rate
Latency
Delta
- RateInput intensity ↔ Amount of firing
- LatencyInput intensity ↔ Timing of firing
- DeltaInput change ↔ Spike occurrence
Trade‑off: accuracy vs energy
Research goals
- Compare LSTM, TCN, SNN‑only, Hybrid TCN–SNN, and SpikingTCN on sEMG gestures.
- Evaluate rate/latency/delta encodings and temporal resolution (Ts).
- Clarify the accuracy–energy trade‑off through firing‑rate/sparsity statistics.