Designing Spiking Neural Controllers for Neuroprosthetic Systems
Junho Park
Supervisor: Dr. Luca Manneschi
MSc Cybersecurity & AI
Background & Motivation
  • Neuroprosthetics → Human–Computer Interaction (HCI)
  • Neuroprosthetic Systems(hand)

    What is a neuroprosthetic system? It is a tool that allows signal information to translate into prosthetic movements.

  • sEMG → Predictive, non‑invasive muscle activity signal
  • Challenges → Noisy signals, high‑dimensional features, ↑ energy cost
⚡ Accuracy vs Battery ⌛ Real‑time 🔋 Low‑power

Why this research? Neuroprosthetic control requires reliable decoding(gesture prediction) under strict power and latency budgets. We study encoding schemes (Rate/Delta/Latency) and model families (LSTM, TCN, SNN, Hybrid, SpikingTCN) to map the accuracy–efficiency trade‑off.

Neuroprosthetic System Flow
Sensors
sEMG electrodes
Preprocess · Encode
Labeling · Rate/Delta/Latency
Model
LSTM / TCN / SNN / Hybrid / SpikingTCN
Actuators
Artificial hand
Objectives