Models of swarming and modes of controlling them are numerous however, to date swarm researchers have mostly ignored a fundamental problem that impedes scalable human interaction with large bio-inspired robot swarms, namely, how do you know what the swarm is doing if you cant observe every agent in the collective Some swarm models exhibit multiple emergent behaviors from the same parameters. This provides increased expressivity at the cost of uncertainty about the swarms actual behavior. Additionally, as robot swarms increase in size, bandwidth and time constraints limit the number of agents that can be controlled and observed. Thus, it is desirable to be able to recognize and monitor the collective behavior of the entire swarm through limited samples of a small subset of agents. We present a novel framework for classifying the collective behavior of a bio-inspired robot swarm using locally based approximations of global features of the emergent collective behaviors. We apply this framework to two bio-inspired models of swarming that exhibit multiple collective behaviors, present a formal metric of expressivity, and develop a classifier that leverages local agent-level features to accurately recognize collective swarm behaviors despite bandwidth limitations.