Statements (48)
| Predicate | Object |
|---|---|
| gptkbp:instance_of |
gptkb:software_framework
|
| gptkbp:application |
gptkb:robot
image recognition natural language processing |
| gptkbp:benefits |
improved efficiency
faster training time reduced labeling costs requires less training data |
| gptkbp:challenges |
overfitting risk
generalization to unseen classes |
| gptkbp:defines |
a type of machine learning where a model learns information about object categories from a single training example.
|
| gptkbp:example |
speech recognition
object detection handwriting recognition face recognition |
| gptkbp:future_plans |
applications in healthcare
improvements in user experience advancements in neural networks enhancements in robotics potential in personalized AI |
| https://www.w3.org/2000/01/rdf-schema#label |
one-shot learning
|
| gptkbp:is_compared_to |
deep learning
traditional machine learning |
| gptkbp:is_evaluated_by |
gptkb:municipality
F1 score accuracy precision Euclidean distance cosine similarity Mahalanobis distance |
| gptkbp:provides_information_on |
gptkb:Stanford_Dogs
gptkb:CUB-200-2011 gptkb:Celeb_A gptkb:Mini_Image_Net gptkb:Omniglot |
| gptkbp:related_to |
transfer learning
few-shot learning |
| gptkbp:research_areas |
gptkb:Artificial_Intelligence
cognitive science computer vision |
| gptkbp:technique |
gptkb:Siamese_networks
data augmentation feature extraction meta-learning memory-augmented neural networks prototypical networks |
| gptkbp:bfsParent |
gptkb:GPT-3
|
| gptkbp:bfsLayer |
4
|