![](/img/trans.png)
[英]How to optimize a CNN in Keras using precision (instead of accuracy)
[英]How to optimize the ratio of (True positive)/(False postive) instead of accuracy?
經典指標是“准確率”,與:(真陽性+真陰性)/(假陽性+假陰性)有關
在分類問題中,假陰性比假陽性更容易容忍。 也就是說,我想分配更多的權重來改善(真陽性)/(假陽性)。 如何做到這一點?
model.compile(optimizer='adam',
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])
Tensorflow allows sensitivities to be shifted for these metrics https://www.tensorflow.org/api_docs/python/tf/keras/metrics/SensitivityAtSpecificity , or if you want the false positives directly (which I think only gives you access to the number誤報(如果有幫助): https://www.tensorflow.org/api_docs/python/tf/keras/metrics/FalsePositives 。 我對 tensorflow 不太了解,但我希望這會有所幫助
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.