LATIDIA · Ciberseguridad

Desaprendizaje automático para modelos de lenguaje grandes: fundamentos, avances y extensiones genéticas

arXiv: 2609.30909v1Tipo de anuncio: nuevo Resumen: El desaprendizaje automático tiene como objetivo eliminar la influencia del objetivo y preservar otras capacidades. Esta encuesta compara métodos, puntos de referencia y evidencia en grandes modelos de lenguaje

La noticia

arXiv:2609.30909v1 Announce Type: new Abstract: Machine unlearning aims to remove target influence while preserving other capabilities. This survey compares methods, benchmarks, and evidence across large language models and systems using retrieval, memory, tools, and interacting agents. A five-layer framework connects removal requests, system boundaries, target locations, interventions, and supported claims. A seven-stage lifecycle and six evidence dimensions guide comparison. The review shows that target construction, retained data, and recovery tests affect reported outcomes. Evidence from model evaluations remains insufficient to establish removal across external state and subsequent updates, motivating evaluation that tracks dependencies and tests whether target influence returns.