Supernovae probably occur about once per century in our own (quite large) galaxy.
If we were to assume similar star formation rates and a similar stellar birth-mass distribution in other galaxies and that the galaxies in the HST-UDF were similar to our own, then we could also assume that the supernova rate was similar.
The UDF was taken over about 4 months, or 1/300 of a century. So one might expect a supernova to occur in about 1/300 galaxies - so about 300 of the 10,000 galaxies observed will have hosted a supernova during the period that the image was built up.
Of course, even if this figure is correct (see below), many of these may have been missed because the galaxies were too far away or the supernovae were obscured by dust, or because they brightened and faded between observations (there was a 6 week gap in the observations). On the other hand, the UDF was imaged later by HST with other instruments, leading to further opportunities to identify supernova candidates.
The calculated number could be awry in many ways. Most galaxies are smaller than the Milky Way, so one might expect fewer supernovae. On the other hand, most of these galaxies are far away and the star formation rate was likely higher in the past, leading to more supernovae. On the other hand, the supernova rate will be higher in some types of galaxies (Starbursts), but lower in others (ellipticals). Without a very detailed demography of the types, metallicity, masses and ages of the galaxies, and a detailed understanding of how the supernova rate varies with these parameters, your question is difficult to answer.