Guided search in attention is multi Pass filtering in machine learning

In guided search, people make use of pre-attentive low-level features in their search for a more complex item. Essentially, guidance based on a somewhat parallel pre-attentive search provides a preliminary filtering mechanism. Similarly, in semantic search and other machine learning tasks, progressively more complex models are employed to sequentially tread through the items involved. This can be associated with sieves of increased granularity. ``