T-SMOTE: Temporal-oriented Synthetic Minority Oversampling Technique for Imbalanced Time Series Classification.
Pu ZhaoChuan LuoBo QiaoLu WangSaravan RajmohanQingwei LinDongmei ZhangPublished in: IJCAI (2022)
Keyphrases
- minority class
- sampling methods
- class imbalance
- majority class
- class imbalanced
- class distribution
- imbalanced datasets
- imbalanced data sets
- imbalanced data
- classification error
- decision boundary
- nearest neighbour
- spatio temporal
- rare events
- support vector machine
- cost sensitive learning
- random sampling
- spatial and temporal
- original data
- training set
- training dataset
- active learning
- sampling algorithm
- training data
- ensemble learning
- imbalanced class distribution
- temporal data
- highly skewed
- temporal constraints
- temporal reasoning
- high dimensionality
- dynamic time warping
- test set
- temporal information
- concept drift
- data sets
- cost sensitive
- base classifiers
- test data
- training samples
- data points
- highly imbalanced
- machine learning