An issue in Neural Networks, wherein a model does not know when its wrong. A cost function is trained to be confident in specific outputs, there is no option to be unsure of an answer.
Deep Neural Rejection is an attempt to fix this.
An issue in Neural Networks, wherein a model does not know when its wrong. A cost function is trained to be confident in specific outputs, there is no option to be unsure of an answer.
Deep Neural Rejection is an attempt to fix this.