InfoNCE Loss

#deep-learning #loss-in-ml #interview #need-review

InfoNCE loss is the improvement over the Contrastive Loss
In InfoNCE, there is one anchor, one positive and multiple negative samples for each instance
It converge faster than the Triplet Loss
In short, it is the Cross Entropy of the positive ones

InfoNCE Loss Formula

$l o s s = - E_{x} (l o g \frac{f (x^{+})}{f (x^{+}) + \sum_{i \in N} f (x_{i}^{-})})$

Related Notes