Relationship Between Nonsmoothness in Adversarial Training, Constraints of Attacks, and Flatness in the Input Space.
Sekitoshi KanaiMasanori YamadaHiroshi TakahashiYuki YamanakaYasutoshi IdaPublished in: IEEE Trans. Neural Networks Learn. Syst. (2024)
Keyphrases
- input space
- training set
- feature space
- input data
- dimensionality reduction
- kernel function
- output space
- high dimensional
- hyperplane
- high dimensional data
- data points
- high dimensional feature space
- k nearest neighbor
- knn
- regression problems
- supervised learning
- kernel matrix
- input patterns
- support vectors
- low dimensional
- training examples
- linearly separable
- receptive fields
- nearest neighbor
- machine learning
- smooth functions
- embedding space
- euclidean space
- small number
- active learning
- pattern recognition
- feature selection