Semantics of Voids within Data: Ignorance-Aware Machine Learning

Operating with ignorance is an important concern of geographical information science when the objective to discover knowledge from imperfect spatial data. Data mining (driven by discovery tools) about processing available (observed, known, and understood) samples data aiming build a model (e.g., classifier) handle that are not yet observed, or understood. These tools traditionally take semantically labeled (known facts) as input for learning. We want challenge indispensability this approach, we suggest considering things other way around. What if task would be follows: how based on semantics our ignorance, i.e., shape “voids” within space? Can improve traditional classification also modeling ignorance? In paper, provide some algorithms visualization zones in two-dimensional spaces design two ignorance-aware smart prototype selection techniques (incremental adversarial) performance nearest neighbor classifiers. present experiments artificial real datasets test concept usefulness discovery.