Using transfer learning and dimensionality reduction techniques to improve generalisability of machine-learning predictions of mosquito ages from mid-infrared spectra

Table 2 Precision, recall, and f1-score of the best deep learning model for classifying mosquito age classes from alternate sources compared to the best standard machine learning algorithm (i.e. XGBoost classifier)

Model name	Age class (Days)	Precision	Recall	f-1 score	No. of samples per age class
MLP-3	1–9	0.98	0.97	0.98	895
MLP-3	10–17	0.97	0.97	0.97	707
XGB-6	1–9	0.98	0.99	0.98	895
XGB-6	10–17	0.98	0.98	0.98	707

*MLP–3: Multi-Layer Perceptron trained with PCA as a dimensionality reduction technique and 2% transfer learning, XGB-6: XGBoost classifier trained with PCA as a dimensionality reduction technique and 2% target population samples used for transfer learning

ISSN: 1471-2105