al_s2: Implementation of Shortest Shortest Path (S2) algorithm
In e-sensing/activelearning: Active Learning Satellite Image Time Series

Description Usage Arguments Value Author(s) References

Semi-Supervised Learning is a set of classification methods that use both labelled and unlabelled samples. The Shortest Shortest Path (S2) algorithm uses a undirected graph of the samples and iteratively removes edges, trying to identify the boundaries of each class in the graph.

This function receives a sits tibble with time series samples and it returns a sits tibble with either the unlabelled samples to be sent to the oracle or the label of each sample.

al_s2(
  samples_tb,
  sim_method = "correlation",
  closest_n = 6,
  mode = "active_learning"
)

`samples_tb`	A sits tibble with both labelled and unlabelled samples (i.e. NA).
`sim_method`	A character. A method for computing the similarity among samples. See proxy::simil for details.
`closest_n`	An integer. The number of most similar samples to keep while building a similarity graph of the samples.
`mode`	A character telling if the functin runs on either the "active_learning" or "semi_supervised_learning" mode.

1 2	A sits tibble with either the samples to be sent to the oracle (mode "semi_supervised_learning", column

Alber Sanchez, alber.ipia@inpe.br

Dasarathy, G., Nowak, R., & Zhu, X. (2015). S2: An efficient graph based active learning algorithm with application to nonparametric classification. Journal of Machine Learning Research, 40(2015), 1–20.

e-sensing/activelearning documentation built on Dec. 20, 2021, 2:21 a.m.