Metric learning by directly minimizing the k-NN training error.
Konstantin ChernoffMarco LoogMads NielsenPublished in: ICPR (2012)
Keyphrases
- knn
- metric learning
- distance function
- k nearest neighbor
- distance metric
- nearest neighbor
- text categorization
- k nearest neighbour
- similarity search
- generalization error
- text classification
- semi supervised
- pairwise
- dimensionality reduction
- learning tasks
- support vector machine svm
- multi task
- euclidean distance
- feature selection
- model selection
- semi supervised learning
- feature space
- support vector machine
- reinforcement learning
- sample size
- learning problems
- learning algorithm
- upper bound
- neural network