One Shot and Metric Learning - Quadruplet Loss
Published:Jun 2, 2020 11:30
•1 min read
•ML Street Talk Pod
Analysis
This article summarizes a podcast episode discussing one-shot learning, metric learning, and quadruplet loss, focusing on Eric Craeymeersch's work. It highlights the shift towards contrastive architectures and mentions related papers and articles.
Key Takeaways
- •The podcast episode covers ML engineering, computer vision, siamese networks, contrastive loss, one-shot learning, and metric learning.
- •The discussion centers around quadruplet loss as a new approach to one-shot learning.
- •The article highlights the use of contrastive architectures like Siamese networks and their evolution from triplet loss to quadruplet loss.
- •Eric Craeymeersch's work and articles are central to the topic.
Reference
“The article references Eric Craeymeersch's Medium articles and the FaceNet paper, providing context for the discussion on quadruplet loss and its application in one-shot learning.”