CNN-based e-nose classification and concentration regression
CuHHTP-5C on micro-LED platform · Electrode
Four-sensor array; EtOH 50/100/200 ppm, TMA 50/100/200 ppm, NH3 10/20/50 ppm, NO2 1/2/5 ppm; sampling rate 1 s; 4 x 60 s sliding window; PyTorch CNN.
| Property | Reported value | Normalised value | Uncertainty | Origin and quality | Source |
|---|---|---|---|---|---|
| Average gas classification accuracyMarked as a best value within this paper | 99.8% | — | — | Text Exact Reported | 7 · Deep learning-based cMOF e-nose system · Fig. 6c |
| CNN gas prediction latency | within 2 min, even considering the 60-s sliding time window | 120 s | upper bound | Text Range | 7 · Deep learning-based cMOF e-nose system · Fig. 6 |
| Mean absolute error for concentration predictionMarked as a best value within this paper | 7.94% | — | — | Text Exact Reported | 7 · Deep learning-based cMOF e-nose system · Fig. 6d |
| TMA 50 ppm concentration prediction MAE | 22.5% | — | — | SI Table Exact Reported | 36 · Supplementary Table 2 · Supplementary Table 2 |