Fig. 2: MNIST and EMNIST only-few-shot classification. | npj Artificial Intelligence

Fig. 2: MNIST and EMNIST only-few-shot classification.

From: Learning from one and only one shot

Fig. 2

a shows sample images from MNIST, and b shows those from EMNIST. The first 1–20 training images per class and the full test set are used. For each classifier, we plot test accuracy versus training size N, as well as the smallest N needed to reach an accuracy threshold (90% for MNIST and 75% for EMNIST due to increased difficulty). Our model outperforms all other comparison models for all N, requiring the least amount of training data to achieve strong performance.

Back to article page