Title: Robust AI Techniques for Large-Scale Surveys of Space and Planetary Science Data Sets Author: P. R. Gazis Space and planetary science research can often be divided into two phases: a 'survey' phase, to identify the objects or events of interest, and an 'analysis' phase, in which these objects or events are analyzed, interpreted, or compared to some set of models. In the past, when data sets were small, surveys could often be accomplished by direct physical examination of the relevant data, and most of the effort in a research program might be directed to the analysis phase. As data sets have grown larger (with sizes in excess of terabytes for collections of hyperspectral images or CMB measurements), this is no longer the case, and the effort required to perform many forms of surveys threatens to become prohibitive. Over the past several years, we have explored a number of simple well-understood AI schemes to perform fast preliminary surveys of extremely large space and planetary science data sets. These methods have shown considerable promise in the particular applications to which they have been applied. They have potential to enable a wide range of research efforts that would otherwise be difficult to accomplish.