In Newtonian physics, force is equal to the rate of change of momentum. In Lagrangian mechanics, we assert a function of position and velocity that governs the physics called the 'Lagrangian', and then 'force' is the partial derivative of the Langrangian with respect to position, and 'momentum' the partial derivative with respect to velocity. It looks a bit like:
$\frac{\partial L}{\partial x}=\frac{d}{dt}\frac{\partial L}{\partial \dot x}$ where $\frac{\partial L}{\partial x}=F$ and $\frac{\partial L}{\partial \dot x}=mv$
But Lagrangian mechanics allows positions $x$ to be expressed in non-inertial and curved coordinate systems (called 'generalised coordinates'), in which the relationship of the partial derivative to velocities in the flat, inertial coordinate system involves some extra terms. In an inertial reference frame (and assuming constant mass), these extra terms disappear and you just differentiate velocity to get acceleration (i.e. $F = ma$). In non-inertial or curved coordinates, you get extra terms on the right hand side, because you're differentiating with respect to something that itself varies. The trick with 'inertial forces' is just to transfer these extra terms over to the left hand side and call them part of the force. We're used to the form of the force being arbitrary, more so than the formula for acceleration.
Newtonian physics asserts as a postulate the existence of a set of inertial reference frames where these curvy extra terms vanish. It's not the only viewpoint on the matter, though.
In General Relativity, we allow much more freedom with regard to coordinate systems. If we consider a rock spinning in a static universe full of distant stars, Mach's principle suggests that this should be equivalent to a static rock in a universe with the distant stars spinning around it. And indeed, General Relativity predicts that there should be a gravitomagnetic force (a component of gravity analogous to the magnetic force in electromagnetism) that arises when there is a flow or current of mass, and in the case of the spinning universe turns out to match the centrifugal force. (See here.) Thus, the supposedly fictitious centrifugal force is just the gravitational force exerted by a spinning universe, and not fictitious at all! (Or at least, no more fictitious than gravity.)
In a sense, General Relativity pulls the same trick in reverse - it transfers all the forces over to the right-hand side of the equation and calls them curvature, leaving no 'real' forces. All particles follow free-fall paths through curved spacetime, and all forces are just aspects of curvature.
Of course, that doesn't apply in Newtonian mechanics, that asserts a simple inverse-square law for gravity, and a God-given set of inertial frames. Forces that can't be made to vanish and remain in the inertial frames are classified as 'real'.