npsm logo

Download original image
Fig. 5. (Color online) (a) is a visualization of t-SNE for the open set. The model was tested similarly to the closed set, by increasing the size of the training data from 10 IDs to 100 IDs. (b) is the confusion matrix visualization in open set trained model and closed set test samples. The results showed that training with relatively small data (10 labels) led to better label consistency.
New Phys.: Sae Mulli 2023;73:385~394