Selected article for: "accuracy achieve and achieve performance"
Author: Jakub M Bartoszewicz; Anja Seidel; Bernhard Y Renard
Title: Interpretable detection of novel human viruses from genome sequencing data
  • Document date: 2020_1_30
    • ID: ac00tai9_47
    Snippet: Adding more diversity to the negative class may still boost performance on more diverse test sets, as in the case of CNN trained on the "All" dataset (CNN All ). This model performs a bit worse on viruses infecting hosts related to humans, but achieves higher accuracy than the "Chordata"-trained models and the best recall overall. Rebalancing the negative class using the "Stratified" dataset helps to achieve higher performance on animal viruses w.....
    Document: Adding more diversity to the negative class may still boost performance on more diverse test sets, as in the case of CNN trained on the "All" dataset (CNN All ). This model performs a bit worse on viruses infecting hosts related to humans, but achieves higher accuracy than the "Chordata"-trained models and the best recall overall. Rebalancing the negative class using the "Stratified" dataset helps to achieve higher performance on animal viruses while maintaing high overall accuracy. The LSTMs are outperformed by the CNNs, but they can be used for shorter reads without retraining (see Sections 2.2 and 3.2).

    Search related documents:
    Co phrase search for related documents
    • CNN dataset and high accuracy: 1, 2, 3, 4, 5, 6, 7, 8, 9
    • CNN dataset and high performance: 1, 2, 3, 4
    • CNN dataset train and high accuracy: 1, 2, 3, 4
    • diverse test set and high accuracy: 1
    • good recall and high accuracy: 1, 2