arXiv subjects

Haik Manukian

Publications and source records attributed to Haik Manukian.

10 recordsLinked to original sources

arXiv2024

Quantum control in artificial neurons with superconductor-ionic memory inserted in the feedback

To improve artificial intelligence/autonomous systems and help with treating neurological conditions, there's a requirement for artificial neuron hardware that mimics biological. We examine experimental artificial neurons with quantum tunneling memory using 4.2 nm of ionic Hafnium oxide and Niobium metal inserted in the positive and negative feedback of an oscillator. These neurons have adaptive spiking behavior and hybrid non-chaotic/chaotic modes. When networked, they output with strong itinerancy. The superconducting state at 8.1 Kelvin results in Josephson tunneling with signs that the ionic states are influenced by quantum coherent control in accordance with quantum master equation calculations of the expectation values and correlation functions with a calibrated time dependent Hamiltonian. We experimentally demonstrate a learning network of 4 artificial neurons, and the modulation of signals.

quant-ph↗

arXiv2021

Mode-Assisted Joint Training of Deep Boltzmann Machines

The deep extension of the restricted Boltzmann machine (RBM), known as the deep Boltzmann machine (DBM), is an expressive family of machine learning models which can serve as compact representations of complex probability distributions. However, jointly training DBMs in the unsupervised setting has proven to be a formidable task. A recent technique we have proposed, called mode-assisted training, has shown great success in improving the unsupervised training of RBMs. Here, we show that the performance gains of the mode-assisted training are even more dramatic for DBMs. In fact, DBMs jointly trained with the mode-assisted algorithm can represent the same data set with orders of magnitude lower number of total parameters compared to state-of-the-art training procedures and even with respect to RBMs, provided a fan-in network topology is also introduced. This substantial saving in number of parameters makes this training method very appealing also for hardware implementations.

cs.LG↗

arXiv2020

Generating Weighted MAX-2-SAT Instances of Tunable Difficulty with Frustrated Loops

Many optimization problems can be cast into the maximum satisfiability (MAX-SAT) form, and many solvers have been developed for tackling such problems. To evaluate a MAX-SAT solver, it is convenient to generate hard MAX-SAT instances with known solutions. Here, we propose a method of generating weighted MAX-2-SAT instances inspired by the frustrated-loop algorithm used by the quantum annealing community. We extend the algorithm for instances of general bipartite couplings, with the associated optimization problem being the minimization of the restricted Boltzmann machine (RBM) energy over the nodal values, which is useful for effectively pre-training the RBM. The hardness of the generated instances can be tuned through a central parameter known as the frustration index. Two versions of the algorithm are presented: the random- and structured-loop algorithms. For the random-loop algorithm, we provide a thorough theoretical and empirical analysis on its mathematical properties from the perspective of frustration, and observe empirically a double phase transition behavior in the hardness scaling behavior driven by the frustration index. For the structured-loop algorithm, we show that it offers an improvement in hardness over the random-loop algorithm in the regime of high loop density, with the variation of hardness tunable through the concentration of frustrated weights.

cs.LG↗

arXiv2020

Mode-Assisted Unsupervised Learning of Restricted Boltzmann Machines

Restricted Boltzmann machines (RBMs) are a powerful class of generative models, but their training requires computing a gradient that, unlike supervised backpropagation on typical loss functions, is notoriously difficult even to approximate. Here, we show that properly combining standard gradient updates with an off-gradient direction, constructed from samples of the RBM ground state (mode), improves their training dramatically over traditional gradient methods. This approach, which we call mode training, promotes faster training and stability, in addition to lower converged relative entropy (KL divergence). Along with the proofs of stability and convergence of this method, we also demonstrate its efficacy on synthetic datasets where we can compute KL divergences exactly, as well as on a larger machine learning standard, MNIST. The mode training we suggest is quite versatile, as it can be applied in conjunction with any given gradient method, and is easily extended to more general energy-based neural network structures such as deep, convolutional and unrestricted Boltzmann machines.

cs.LG↗

arXiv2019

Data Generation for Neural Programming by Example

Programming by example is the problem of synthesizing a program from a small set of input / output pairs. Recent works applying machine learning methods to this task show promise, but are typically reliant on generating synthetic examples for training. A particular challenge lies in generating meaningful sets of inputs and outputs, which well-characterize a given program and accurately demonstrate its behavior. Where examples used for testing are generated by the same method as training data then the performance of a model may be partly reliant on this similarity. In this paper we introduce a novel approach using an SMT solver to synthesize inputs which cover a diverse set of behaviors for a given program. We carry out a case study comparing this method to existing synthetic data generation procedures in the literature, and find that data generated using our approach improves both the discriminatory power of example sets and the ability of trained machine learning models to generalize to unfamiliar data.

cs.LG↗

arXiv2018

Accelerating Deep Learning with Memcomputing

Restricted Boltzmann machines (RBMs) and their extensions, called 'deep-belief networks', are powerful neural networks that have found applications in the fields of machine learning and artificial intelligence. The standard way to training these models resorts to an iterative unsupervised procedure based on Gibbs sampling, called 'contrastive divergence' (CD), 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 an efficient alternative to CD by means of simulations of digital memcomputing machines (DMMs). We test our approach on pattern recognition using a modified version of the MNIST data set. DMMs sample effectively the vast phase space given by the model distribution of the RBM, and provide a very good approximation close to the optimum. This efficient search significantly reduces the number of pretraining iterations necessary to achieve a given level of accuracy, as well as a total performance gain over CD. In fact, the acceleration of pretraining achieved by simulating DMMs is comparable to, in number of iterations, the recently reported hardware application of the quantum annealing method on the same network and data set. Notably, however, DMMs perform far better than the reported quantum annealing results in terms of quality of the training. We also compare our method to advances in supervised training, like batch-normalization and rectifiers, that work to reduce the advantage of pretraining. We find that the memcomputing method still maintains a quality advantage ($>1\%$ in accuracy, and a $20\%$ reduction in error rate) over these approaches. Furthermore, our method is agnostic about the connectivity of the network. Therefore, it can be extended to train full Boltzmann machines, and even deep networks at once.

cs.LG↗

arXiv2018

Instantons in self-organizing logic gates

Self-organizing logic is a recently-suggested framework that allows the solution of Boolean truth tables "in reverse," i.e., it is able to satisfy the logical proposition of gates regardless to which terminal(s) the truth value is assigned ("terminal-agnostic logic"). It can be realized if time non-locality (memory) is present. A practical realization of self-organizing logic gates (SOLGs) can be done by combining circuit elements with and without memory. By employing one such realization, we show, numerically, that SOLGs exploit elementary instantons to reach equilibrium points. Instantons are classical trajectories of the non-linear equations of motion describing SOLGs, and connect topologically distinct critical points in the phase space. By linear analysis at those points, we show that these instantons connect the initial critical point of the dynamics, with at least one unstable direction, directly to the final fixed point. We also show that the memory content of these gates only affects the relaxation time to reach the logically consistent solution. Finally, we demonstrate, by solving the corresponding stochastic differential equations, that since instantons connect critical points, noise and perturbations may change the instanton trajectory in the phase space, but not the initial and final critical points. Therefore, even for extremely large noise levels, the gates self-organize to the correct solution. Our work provides a physical understanding of, and can serve as an inspiration for, new models of bi-directional logic gates that are emerging as important tools in physics-inspired, unconventional computing.

cs.ET↗

arXiv2017

Memcomputing Numerical Inversion with Self-Organizing Logic Gates

We propose to use Digital Memcomputing Machines (DMMs), implemented with self-organizing logic gates (SOLGs), to solve the problem of numerical inversion. Starting from fixed-point scalar inversion we describe the generalization to solving linear systems and matrix inversion. This method, when realized in hardware, will output the result in only one computational step. As an example, we perform simulations of the scalar case using a 5-bit logic circuit made of SOLGs, and show that the circuit successfully performs the inversion. Our method can be extended efficiently to any level of precision, since we prove that producing n-bit precision in the output requires extending the circuit by at most n bits. This type of numerical inversion can be implemented by DMM units in hardware, it is scalable, and thus of great benefit to any real-time computing application.

cs.ET↗

arXiv2014

PS1-10jh: The Disruption of a Main-Sequence Star of Near-Solar Composition

When a star comes within a critical distance to a supermassive black hole (SMBH), immense tidal forces disrupt the star, resulting in a stream of debris that falls back onto the SMBH and powers a luminous flare. In this paper, we perform hydrodynamical simulations of the disruption of a main-sequence star by a SMBH to characterize the evolution of the debris stream after a tidal disruption. We demonstrate that this debris stream is confined by self-gravity in the two directions perpendicular to the original direction of the star's travel, and as a consequence has a negligible surface area and makes almost no contribution to either the continuum or line emission. We therefore propose that any observed emission lines are not the result of photoionization in this unbound debris, but are produced in the region above and below the forming elliptical accretion disk, analogous to the broad-line region (BLR) in steadily-accreting active galactic nuclei. As each line within a BLR is observationally linked to a particular location in the accretion disk, we suggest that the absence of a line indicates that the accretion disk does not yet extend to the distance required to produce that line. This model can be used to understand the spectral properties of the tidal disruption event (TDE) PS1-10jh, for which He II lines are observed, but the Balmer series and He I are not. Using a maximum likelihood analysis, we show that the disruption of a main-sequence star of near-solar composition can reproduce this event.

astro-ph.HE↗

arXiv2014

Turbovelocity Stars: Kicks Resulting From the Tidal Disruption of Solitary Stars

The centers of most known galaxies host supermassive black holes (SMBHs). In orbit around these black holes are a centrally-concentrated distribution of stars, both in single and in binary systems. Occasionally, these stars are perturbed onto orbits that bring them close to the SMBH. If the star is in a binary system, the three-body interaction with the SMBH can lead to large changes in orbital energy, depositing one of the two stars on a tightly-bound orbit, and its companion into a hyperbolic orbit that may escape the galaxy. In this Letter, we show that the disruption of solitary stars can also lead to large positive increases in orbital energy. The kick velocity depends on the amount of mass the star loses at pericenter, but not on the ratio of black hole to stellar mass, and are at most the star's own escape velocity. We find that these kicks are usually too small to result in the ejection of stars from the Milky Way, but can eject the stars from the black hole's sphere of influence, reducing their probability of being disrupted again. We estimate that ~10^5 stars, ~1% of all stars within 10 pc of the galactic center, are likely to have had mass removed by the central black hole through tidal interaction, and speculate that these "turbovelocity" stars will at first be redder, but eventually bluer, and always brighter than their unharrassed peers.

astro-ph.HE↗