ENERGY EFFICIENT PROBABILISTIC SOLUTIONS FOR BEYOND-ISING APPLICATIONS

<p dir="ltr">This research applies probabilistic computing, a novel sub-field in computing algorithms, to various problems in an effort to grow the application space of a field that has largely been restricted to ising-style optimization and sampling problems. These probabilistic approaches generally improve time-to-solution, decrease energy usage, or expand on the capabilities of the existing deterministic approach in meaningful ways. </p><p dir="ltr">Probabilistic computing is established as a novel paradigm of taking advantage of samples over probabilistic variables to achieve system-level benefits via intelligently designed hardware. This research works across the stack, tackling applications that have not been studied before in a probabilistic context, presenting novel algorithms for those applications, and developing novel hardware architectures to implement those algorithms efficiently. </p>