Similarity search

The aeon.similarity_search module contains algorithms and tools for similarity search tasks. First, we distinguish between series estimator and collection estimators, similarly to the aeon.transformer module. Secondly, we distinguish between estimators used neighbors (with suffix SNN for subsequence nearest neighbors, or ANN for approximate nearest neighbors) search and estimators used for motifs search.

Series Similarity search estimators

DummySNN(length[, normalize, n_jobs])	Estimator to compute the on profile and distance profile using brute force.
MassSNN(length[, normalize])	Estimator to compute the subsequences nearest neighbors using MASS _[1].

StompMotif(length[, normalize])

Estimator to extract top k motifs using STOMP, described in [R498399f961be-1].

Collection Similarity search estimators

RandomProjectionIndexANN([n_hash_funcs, ...])

Random Projection Locality Sensitive Hashing index with cosine similarity.

Base Estimators

BaseSimilaritySearch()

Base class for similarity search applications.

BaseSeriesSimilaritySearch([axis])

Base class for similarity search applications on single series.

BaseCollectionSimilaritySearch()

Similarity search base class for collections.