TCN–SNN for sEMG Gesture
The research investigates spiking-based neural architectures for surface electromyography (sEMG)–driven neuroprosthetic control. Composite score over spikes;
I build efficiency-first models around Spiking Neural Networks (SNNs) and apply them to Robotics (ROS2), Security RAG, and On-device ML. The goal: ship AI that meets real-world constraints: latency, cost, spikes/power.
Case studies with metrics, evaluation, and deployment notes.
The research investigates spiking-based neural architectures for surface electromyography (sEMG)–driven neuroprosthetic control. Composite score over spikes;
A production-ready template for SNN/TCN-based edge robotics projects (ROS2 + snnTorch + FastAPI + Docker). Focus: latency, spike/energy metrics, edge deployment (Pi/Jetson).
Embedding & prompt experiments to reduce hallucination, cost, and latency. Automatic evaluation dashboard with a FastAPI service.
Open to Research Engineer (Neuromorphic/Robotics), Applied ML (Security), and Robotics + ML roles.
Email —
ghdlwnsgh25@gmail.com
GitHub —
github.com/parkjunho12
LinkedIn —
/in/park-junho-881222164/
CV —
Download PDF
Dissertation —
Download PDF
Dissertation slides —
Open HTML