Transfer learning based few-shot classification using optimal transport mapping from preprocessed latent space of backbone neural network.
Tomás ChobolaDaniel VasataPavel KordíkPublished in: CoRR (2021)
Keyphrases
- transfer learning
- latent space
- neural network
- text classification
- learning tasks
- machine learning algorithms
- machine learning
- knowledge transfer
- feature space
- pattern recognition
- labeled data
- latent variables
- reinforcement learning
- cross domain
- feature extraction
- collaborative filtering
- dimensionality reduction
- manifold learning
- text categorization
- multi task
- feature vectors
- high dimensional
- supervised learning
- gaussian process
- low dimensional
- semi supervised learning
- gaussian processes
- knn
- active learning
- training set
- support vector
- feature selection
- text mining
- semi supervised
- support vector machine
- data analysis
- decision trees
- training samples
- k nearest neighbor
- image classification
- parameter space
- matrix factorization
- probabilistic model
- image retrieval
- learning algorithm
- lower dimensional
- data mining