Jan 01, 2018

Accelerating Deep Learning with Memcomputing

Fabio Lorenzo Traversa, Massimiliano Di Ventra, Haik Manukian

---

Restricted Boltzmann machines (RBMs) and their extensions, often called “deep-belief networks”, are very powerful neural networks that have found widespread applicability in the fields of machine learning and big data. The standard way to training these models resorts to an iterative unsupervised procedure based on Gibbs sampling, called “contrastive divergence”, and additional supervised tuning via back-propagation. However, this procedure has been shown not to follow any gradient and can lead to suboptimal solutions. In this paper, we show a very efficient alternative to contrastive divergence by means of simulations of digital memcomputing machines (DMMs). We test our approach on pattern recognition using the standard MNIST data set of hand-written numbers…

Go To Publication  →

View arXiv version  →

• All Tags Absence of periodic orbits in digital memcomputing machines with solutions Accelerating Deep Learning with Memcomputing Aircraft Loading Optimization: MemComputing the 5th Airbus Problem Coupled oscillator networks for von Neumann and non von Neumann computing Digital Memcomputing: from Logic to Dynamics to Topology Directed percolation and numerical stability of simulations of digital memcomputing machines Dynamic computing random access memory: A brain-inspired computing paradigm with memelements Efficient solution of Boolean satisfiability problems with digital memcomputing Evidence of an exponential speed-up in the solution of hard optimization problems Global minimization via classical tunneling assisted by collective force field formation Instantons in Self-Organizing Logic Gates MemComputing Integer Linear Programming Memcomputing Numerical Inversion With Self-Organizing Logic Gates Memcomputing: Leveraging memory and physics to compute efficiently Mode-Assisted Unsupervised Learning of Restricted Boltzmann Machines On the Universality of Memcomputing Machines Polynomial-time solution of prime factorization and NP-complete problems with digital memcomputing machines Stress-testing memcomputing on hard combinatorial optimization problems Taming a non-convex landscape with dynamical long-range order: memcomputing the Ising spin-glass Topological Field Theory and Computing with Instantons