A recent social media post from Andrew Ng: "Deep Learning is getting really good on Big Data/millions of images. But Small Data is important too. Am seeing many exciting applications at Landing AI where you can get good results with 100 images. Hope more researchers work on Small Data--ML needs more innovations there."
So, what are the currently available methodologies to learn from small data (without overfitting)?
Liming Yuan
Hello,
They are called 'fewshot learning' . The following methods are proposed to solve the small sample problems:
1.Baseline nearest neighbors
2.Matching network
3.Prototypical network
4.Graph neural network
5.Meta-learning LSTM
6.Model-Agnostic
1 votes
1 thanks
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining
More Abdullah Al Imran's questions See All
Similar questions and discussions