oh-yu/domain-invariant-learning

wrongly calculated t-SNE

Closed this issue · 1 comments

✗
    target_feature_tsne = tsne.fit_transform(target_feature)
    source_feature_tsne = tsne.fit_transform(source_feature)
○
N_target = target_feature.shape[0]
feature_tsne = tsne.fit_transform(np.concatenate([target_feature, source_feature], axis=0))
target_feature_tsne, source_feature_tsne = feature_tsne[:N_target], feature_tsne[N_target:]

resolved by b28ab58