Hard negative mining
Hard negative mining, also referred to as hard negative sampling, is a family of techniques in machine learning for selecting or constructing negative samples more difficult to distinguish from a corresponding positive example with respect to uniformly-sampled negatives, leading to a more informative training signal for discriminative classifiers and similarity models.
Overview schema of hard negative mining from data. In the example above, the model is trained to classify gray cats from anything else. The "Hard Negative Selector" prioritizes other images of cats with a different fur color, since they are the most difficult negatives to distinguish from the positive samples. Hard negative mining, also referred to as hard negative sampling, is a family of techniques in machine learning for selecting or constructing negative samples more difficult to distinguish from a corresponding positive example with respect to uniformly-sampled negatives, leading to a…
saved by
related reading
- The Little Book of Deep Learningfleuret.org
- Support vector machine - Wikipediaen.wikipedia.org
- Competing with sampling — Alignment Research Centeralignment.org
- Untangling the Moons: A Visual History of Contrastive Learningtahabouhsine.com
- Contrastive Representation Learning | Lil'Loglilianweng.github.io
- Word2Vec Tutorial Part 2 - Negative Sampling · Chris McCormickmccormickml.com
- Thinking about High-Quality Human Data | Lil'Loglilianweng.github.io
- d89a66c7c80a29b1bdbab0f2a1a94af8-Paper.pdfproceedings.neurips.cc
- deeplearningbook.org/contents/ml.htmldeeplearningbook.org
- ARC progress update: Competing with sampling — LessWronglesswrong.com
- A Course in Machine Learningciml.info
- Support vector machine - Wikipediaen.wikipedia.org