Nowadays, test and measurement tasks at high volume production facilities are fully automated. Normally the responses of analog components to test stimuli have to be first digitized before being automatically processed in order to identify deviations from the reference signal. When dealing with high frequency devices, the analog to digital conversion process becomes costly and/or involves data losses. This situation becomes much more critical when measurement equipment has to become a part of the system itself (BIST). A novel testing technique that avoids excessive costs is proposed in this paper. It is based on the ability of artificial neural networks to classify objects and phenomena and detect deviations from expected results. Our approach is analog to digital data conversion-independent and thus can target high-frequency continuous-time signals.