Accuracy Improvement of Localization and Mapping of ICP-SLAM via Competitive Associative Nets and Leave-One-Out Cross-Validation.
Shuichi KurogiYoichiro YamashitaHikaru YoshikawaKotaro HirayamaPublished in: ICONIP (2) (2014)
Keyphrases
- leave one out cross validation
- loop closing
- model selection
- variable selection
- simultaneous localization and mapping
- support vector machine
- k nearest neighbour
- mobile robot
- generalization error
- cross validation
- confidence intervals
- kernel learning
- closed form
- feature subset
- neural network
- knn
- training data
- information gain
- error rate
- kernel function
- classification accuracy
- kernel parameters
- learning algorithm
- machine learning