Skip to main content

Table 5 BoW vs Sparse coding for top 10 terms in all stages

From: Image-level and group-level models for Drosophilagene expression pattern annotation

Stages BoW Sparse coding
  Accuracy AUC Sensitivity Specificity Accuracy AUC Sensitivity Specificity
4-6 76.27 0.7345 0.6891 0.7798 77.35 0.7437 0.7002 0.7872
7-8 78.47 0.7635 0.7006 0.8264 78.42 0.7666 0.7209 0.8123
9-10 74.28 0.7333 0.6977 0.7669 78.98 0.7592 0.7481 0.7702
11-12 80.14 0.7974 0.7946 0.8002 79.84 0.8021 0.8153 0.7889
13-17 81.02 0.7869 0.7412 0.8325 82.04 0.8012 0.7655 0.8359
  1. Spatial sparse coding performs slightly better than spatial BOW (image-level).