The UV (from young stars) and IR (from dust) emissions in galaxies are
strongly physically linked: UV photons from young stars heat very
efficiently the dust which re-emits the absorbed energy in IR. Therefore
both quantities are star formation tracers but they are anti-correlated
by the process of dust attenuation.
Measuring reliable star formation rates and having robust measurements
of dust attenuation is only possible when both quantities (UV and IR
emissions) are available. However, even in this case, the selection
applied to the sample can strongly modify the results.
In this talk I will report results on our analyses of UV and IR selected
samples of galaxies from z=0 to z~2, based on Herschel and AKARI
surveys. I will show that dust attenuation is much higher in an IR
selection than in a UV one and that both have a similar evolution with
redshift.
I will propose a simple explanation of these results if the stellar mass
is considered as the main driver for dust attenuation.