Set expansion aims to expand a small set of seed entities into a complete set of relevant entities. Most existing approaches assume the input seed set is unambiguous and completely ignore the multi-faceted semantics of seed entities. As a result, given the seed set {"Canon", "Sony", "Nikon"}, previous methods return one mixed set of entities that are either Camera Brands or Japanese Companies. In this paper, we study the task of multi-faceted set expansion, which aims to capture all semantic facets in the seed set and return multiple sets of entities, one for each semantic facet. We propose an unsupervised framework, FUSE, which consists of three major components: (1) facet discovery module: identifies all semantic facets of each seed entity by extracting and clustering its skip-grams, and (2) facet fusion module: discovers shared semantic facets of the entire seed set by an optimization formulation, and (3) entity expansion module: expands each semantic facet by utilizing an iterative algorithm robust to skip-gram noise. Extensive experiments demonstrate that our algorithm, FUSE, can accurately identify multiple semantic facets of the seed set and generate quality entities for each facet.