The inverse relation of a bijective binary function is a 2-valued function, a multivalued function. If this inverse relation should be an algebraic function, it has to be a function. But a function cannot be multivalued. A 2-valued function is no function. "2-valued function" and "multivalued function" are misnomers because a function can be only one-to-one or many-to-one. For treating multivalued functions as functions, you have to treat them e.g. as set-valued functions or as tuple-valued functions.
An algebraic function $f$ with $y=f(x_1,x_2,...,x_n)$ is by definition a root of a polynomial equation (the defining algebraic equation of $f$)
$$p_0(x_1,...,x_n)+p_1(x_1,...,x_n)y+p_2(x_1,...,x_n)y^2+...+p_n(x_1,...,x_n)y^n=0$$
where the $p_i$ ($\forall i=0,1,2,...,n$) are polynomials in $x_1,x_2,...,x_n$. If you want to allow $f$ to be set-valued or tuple-valued, you have to allow powers of sets (respectively powers of tuples) for $y^2,y^3,...,y^n$ in the defining algebraic equation of $f$.