Write the triplet loss formula: max(d(a,p) - d(a,n) + margin, 0)

Can you tell friends apart even if they've changed a lot over time?

Imagine you have a photo album with pictures of your friends. Over time, some friends change their hairstyle, gain weight, or even grow older. You want to easily recognize them even if they look different.

Think of your friends' photos as points in a space. Triplet loss helps you place them closer together if they're the same person and farther apart if they're different people. It's like drawing invisible lines connecting friends' photos.

Example

You have three photos: A, B, and C. Photo A shows your friend Tom in his 20s, photo B shows Tom in his 30s, and photo C shows Tom in his 40s. Triplet loss helps you understand that photos A, B, and C are closer together than photo C and any other photo of a different person.

Remember this

Triplet loss ensures that photos of the same person stay close together, making it easier to recognize them even if they change over time.

Text adapted from Wikipedia, licensed under CC BY-SA 4.0.

Related concepts

Write the contrastive loss function for SimCLR

Contrastive loss function: L = (1/2N) Σ [max(0, margin - y_i * (z_i - z_j))^2 + max(0, y_i * (z_i - z_j) - margin)^2]

Cross-entropy

Cross-entropy loss equation: H(p, q) = -Σ(p(x) * log(q(x)))

PageRank

PageRank formula: PR(A) = (1-d) + d Σ(PR(C)/L(C))

Bayes' theorem

Bayes' theorem formula: P(A|B) = [P(B|A) * P(A)] / P(B)

Jensen–Shannon divergence

Jensen-Shannon divergence formula: D_JS(P||Q) = 1/2 * D_KL(P||(M)) + 1/2 * D_KL(Q||(M))

Distance transform

Manhattan distance formula: |x1 - x2| + |y1 - y2|

Swipe through 100 ML concepts daily

Open Pocket Polymath