Projects
Research prototypes, open-source tools, and applied systems.
SNN-GNN Predictive Coding for MRI Representation Learning
Semi-supervised MRI framework using SNN-GNN gating with a SpikeFormer backbone to fuse supervised and unsupervised objectives for label-efficient predictive coding.
- Improved representation learning under limited annotations and targeted edge-efficient inference.
- Reduced compute / energy by 90% relative to the dense baseline.
EvCorrupt-Bench: Runtime Corruption Detection for Event Cameras
A benchmark and monitor study for detecting sensor corruption in event-camera streams at runtime: six physically modeled failure modes (leak-event hot pixels, mains flicker, activity bursts, dead regions, readout congestion, polarity faults) injected as timed episodes into 42 recordings across four sensor families, with leakage-safe injection and false-alarm-budgeted evaluation. In submission.
- Grew out of a self-supervised SNN-GNN event denoiser (97% reconstruction accuracy), which now serves as label source and fallback action.
- Injection pipeline released as the open-source `evlab` package (pip install evlab): reproducible recipes, per-event ground truth, clock-quantum snapping.
SpikeGuard-RC: Spiking Reservoir Safety Monitoring (in progress)
An always-on spiking reservoir that maintains a compact temporal memory over windowed event-stream features and predicts safe / unsafe / uncertain, gating downstream perception: continue, abstain, or fall back to denoising. Targeting a computer-vision venue submission.
- Corruption-injected freeway benchmark with exact onsets for AUROC and time-to-detection; UAV streams as real-world transfer test.
- Compared against rule-based, MLP, GRU, LSTM, and non-spiking ESN baselines on detection quality, RAM, latency, and spike-rate energy proxy.
Cosmic-Ray Edge AI on CosmicWatch / CREDO Networks
Exploratory edge-AI pipeline over muon-detector networks: mining coincident CosmicWatch events from the CREDO Elasticsearch archive, building event features, and prototyping tiny MLP / SNN classifiers and a simulation-based GNN for multi-detector fusion, toward federated learning across distributed low-cost detectors.
- Data-readiness tooling for detector uptime, gaps, and coincidence mining across the network archive.
- On-detector-scale models: tiny MLP and pure-PyTorch SNN baselines; GNN prototype for synchronized multi-node fusion.
Content-Adaptive Neuromorphic Semantic Communication
SpikeAdapt-SC, an SNN-based semantic communication framework for aerial edge AI.
- 32–48% estimated computation energy savings via SynOps over MAC-based encoding.
- Reduced payload from 64 KB to 48 KB per image (25% bandwidth savings) with zero accuracy penalty at ρ = 0.75.
- Achieved 96.2% adaptive BSC accuracy vs. 95.8% fixed full-rate; >93% accuracy under severe conditions (BER = 0.3, SNR = −2 dB).
PokeAgent Speedrun
Open-source Pokémon Emerald speedrunning agent with 4 agent scaffolds (four-module, simple, ReAct, ClaudePlays) and a multiprocess emulator/agent architecture for stable long-horizon execution.
- Documented lightweight simple pipeline runs 3–5× faster and retains the last 20 actions for compact action history.
- Production-oriented execution paths including no-ocr mode and automated gameplay recording at 30 FPS.