ERS-2026-001Engineering Readiness SeriesReleased
INT8 Post-Training Quantization for Seq2Point NILM on Resource-Constrained Edge Platforms
An engineering assessment of INT8 post-training quantization applied to a Seq2Point NILM model for resource-constrained Edge AI platforms. Parameter storage, behaviour, accuracy and numerical stability were measured. Hardware execution latency, processor utilisation, energy consumption and hardware deployment were not evaluated.
0%
Parameter Storage Reduction
0.0000
Pearson Correlation
0.00 KB
INT8 Parameter Storage
- Framework
- MATLAB Deep Learning Toolbox
- Dataset
- UK-DALE
- Target Appliance
- Water Heater (Boiler)
- Architecture
- Seq2Point
- Optimization
- INT8 post-training quantization
- Deployment
- Resource-Constrained Edge AI
Does not evaluate
- Hardware execution latency
- Energy consumption
- Processor utilisation
- Hardware-specific optimisation
- Cross-dataset generalisation
- Hardware-in-the-loop evaluation
- Long-term operational performance
Publication Details
- Report
- ERS-2026-001 · Radianode Engineering Readiness Series
- Author
- Oluwanifemi A. Ogunjemilua · ORCID
- Issued by
- Radianode Ltd
Related Publications
Radianode is an independent engineering firm providing technical validation and assessment for connected, embedded and Edge AI systems. If you are making a deployment decision that depends on this analysis, talk to an engineer.