The lack of adequate training data is one of the major hurdles in WiFi-based\nactivity recognition systems. In this paper, we propose Wi-Fringe, which is a\nWiFi CSI-based device-free human gesture recognition system that recognizes\nnamed gestures, i.e., activities and gestures that have a semantically\nmeaningful name in English language, as opposed to arbitrary free-form\ngestures. Given a list of activities (only their names in English text), along\nwith zero or more training examples (WiFi CSI values) per activity, Wi-Fringe\nis able to detect all activities at runtime. In other words, a subset of\nactivities that Wi-Fringe detects do not require any training examples at all.\n