Literature and novelty positioning
Every reference below was fetched/verified from an authoritative source (UCI, ACM, IEEE, NeurIPS proceedings) before being cited — none are invented. Structured source: data/literature/references.json.
Gas Sensor Array Drift Dataset
Vergara, A. (2012) — UCI Machine Learning Repository
Primary dataset for this project. All chronological batches, feature counts, and class labels used on this site are verified against this source.
A survey on concept drift adaptation
Gama, J., Zliobaite, I., Bifet, A., Pechenizkiy, M., Bouchachia, A. (2014) — ACM Computing Surveys, 46(4), Article 44
Foundational survey of concept-drift adaptation strategies; frames why this project treats chronological (not IID) evaluation as the primary protocol.
Novelty consideration: Establishes the general concept-drift vocabulary but does not address electronic-nose sensing, TinyML deployment, or explanation stability — this project's combination remains distinct.
A Unified Approach to Interpreting Model Predictions
Lundberg, S. M., Lee, S.-I. (2017) — Advances in Neural Information Processing Systems 30 (NeurIPS 2017), pp. 4766-4777
Reference explainer against which any resource-aware / proxy explanation strategy in this project would be evaluated for fidelity (planned, not yet executed).
Novelty consideration: SHAP itself is not resource-aware and is not evaluated here for on-device / TinyML deployment cost, drift-stability, or physical energy — this project's resource-aware angle is distinct.
"Why Should I Trust You?": Explaining the Predictions of Any Classifier
Ribeiro, M. T., Singh, S., Guestrin, C. (2016) — Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD 2016)
Alternative local-explanation baseline relevant to the planned fidelity/stability comparison for resource-aware explanations.
Novelty consideration: Same as SHAP: not evaluated for TinyML resource constraints or drift-stability in the original work.
TinyML-Enabled Frugal Smart Objects: Challenges and Opportunities
Sanchez-Iborra, R., Skarmeta, A. F. (2020) — IEEE Circuits and Systems Magazine, 20(3), pp. 4-18
Survey framing for the TinyML deployment constraints (Flash, SRAM, latency, energy) this project targets for the nRF52840 / Cortex-M4F stage.
Novelty consideration: Surveys general TinyML deployment, not sensor-drift-aware models or on-device explanation cost — this project's drift + XAI + hardware combination is not covered.
Novelty matrix
Marks reflect only what each cited work actually demonstrates — this project's own row is marked honestly by current evidence state (see Pipeline), not by aspiration. Cells for unexecuted stages are intentionally unmarked.
| Paper | Dataset | Chronological eval. | Sensor drift | TinyML | XAI | On-device XAI | Fidelity | Stability | Physical latency | Flash/SRAM | Physical energy | PPK2 | Reproducibility |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Gama et al. 2014 (concept drift survey) | General (survey) | ✓ | — | — | — | — | — | — | — | — | — | — | — |
| Lundberg & Lee 2017 (SHAP) | General (tabular/image) | — | — | — | ✓ | — | — | — | — | — | — | — | — |
| Ribeiro et al. 2016 (LIME) | General (tabular/text/image) | — | — | — | ✓ | — | — | — | — | — | — | — | — |
| Sanchez-Iborra & Skarmeta 2020 (TinyML survey) | General (survey) | — | — | ✓ | — | — | — | — | ✓ | ✓ | — | — | — |
| This work (proposed) | UCI Gas Sensor Array Drift | ✓ | ✓ | ✓ | ✓ | — | — | — | — | — | — | — | ✓ |