Fig. 4From: Using transfer learning and dimensionality reduction techniques to improve generalisability of machine-learning predictions of mosquito ages from mid-infrared spectraCNN generalisation and prediction of mosquito age using data from a single insectary (Ifakara) with no dimensionality reduction. A Training and validation classification accuracy for mosquito age classes improved from ~ 60 to 95% as training iterations increased (200 epochs). B A normalised confusion matrix displaying the proportions of correct mosquito age class predictions achieved on the held-out Ifakara data (test set) during model training. C Proportions of correct mosquito age class predictions based on unseen data from the alternate insectary (Glasgow)Back to article page