1: \begin{abstract}

2: It is shown that the renormalization group turns to be a symmetry group in a theory 

3: initially formulated in a space of scale-dependent functions, i.e, those 

4: depending on both the position $x$ and the resolution $a$. Such theory, earlier 

5: described in \cite{Altaisky2010PRD,AK2013}, is finite by construction. The space of 

6: scale-dependent functions $\{ \phi_a(x) \}$ is more relevant to physical reality than 

7: the space of square-integrable functions $\mathrm{L}^2(\R^d)$, because due to the 

8: Heisenberg uncertainty principle, what is really measured in any experiment is 

9: always defined in a region rather than point. The effective action $\Gamma_{(A)}$ of our theory turns to be complementary to 

10: the exact renormalization group effective action. The role of the regulator is played 

11: by the basic wavelet -- an ''aperture function'' of a measuring device used to produce 

12: the snapshot of a field $\phi$ at the point $x$ with the resolution $a$. The standard RG 

13: results for $\phi^4$ model are reproduced.     

14: \end{abstract}

15: