While Boolean logic has been the backbone of information processing, there are computationally hard problems like optimization and associative computing wherein this conventional paradigm is fundamentally inadequate. This results in computational inefficacy, and motivates us to explore new pathways to their solution. In this talk, we introduce an experimental testbed comprising of compact coupled relaxation oscillator based dynamical system that exploits the insulator-metal transition in the correlated material, vanadium dioxide (VO2), to efficiently solve the approximate match between stored and input patterns. Our work is inspired by the understanding that associative computing finds a natural analogue in the energy minimization processes of parallel, coupled dynamical systems. Our work not only elucidates a physics-based computing method but also presents opportunities for building customized analog coprocessors for solving computationally hard problems efficiently.