An official website of the United States government
Abstract Employing the probabilistic nature of unstable nano-magnet switching has recently emerged as a path towards unconventional computational systems such as neuromorphic or Bayesian networks. In this letter, we demonstrate proof-of-concept stochastic binary operation using hard axis initialization of nano-magnets and control of their output state probability (activation function) by means of input currents. Our method provides a natural path towards addition of weighted inputs from various sources, mimicking the integration function of neurons. In our experiment, spin orbit torque (SOT) is employed to “drive” nano-magnets with perpendicular magnetic anisotropy (PMA) -to their metastable state, i.e. in-plane hard axis. Next, the probability of relaxing into one magnetization state (+mi) or the other (−mi) is controlled using an Oersted field generated by an electrically isolated current loop, which acts as a “charge” input to the device. The final state of the magnet is read out by the anomalous Hall effect (AHE), demonstrating that the magnetization can be probabilistically manipulated and output through charge currents, closing the loop from charge-to-spin and spin-to-charge conversion. Based on these building blocks, a two-node directed network is successfully demonstrated where the status of the second node is determined by the probabilistic output of the previous node and a weighted connection between them. We have also studied the effects of various magnetic properties, such as magnet size and anisotropic field on the stochastic operation of individual devices through Monte Carlo simulations of Landau Lifshitz Gilbert (LLG) equation. The three-terminal stochastic devices demonstrated here are a critical step towards building energy efficient spin based neural networks and show the potential for a new application space.
more »
« less
Hardware implementation of Bayesian network building blocks with stochastic spintronic devices
Abstract Bayesian networks are powerful statistical models to understand causal relationships in real-world probabilistic problems such as diagnosis, forecasting, computer vision, etc. For systems that involve complex causal dependencies among many variables, the complexity of the associated Bayesian networks become computationally intractable. As a result, direct hardware implementation of these networks is one promising approach to reducing power consumption and execution time. However, the few hardware implementations of Bayesian networks presented in literature rely on deterministic CMOS devices that are not efficient in representing the stochastic variables in a Bayesian network that encode the probability of occurrence of the associated event. This work presents an experimental demonstration of a Bayesian network building block implemented with inherently stochastic spintronic devices based on the natural physics of nanomagnets. These devices are based on nanomagnets with perpendicular magnetic anisotropy, initialized to their hard axes by the spin orbit torque from a heavy metal under-layer utilizing the giant spin Hall effect, enabling stochastic behavior. We construct an electrically interconnected network of two stochastic devices and manipulate the correlations between their states by changing connection weights and biases. By mapping given conditional probability tables to the circuit hardware, we demonstrate that any two node Bayesian networks can be implemented by our stochastic network. We then present the stochastic simulation of an example case of a four node Bayesian network using our proposed device, with parameters taken from the experiment. We view this work as a first step towards the large scale hardware implementation of Bayesian networks.
more »
« less
- Award ID(s):
- 1739635
- PAR ID:
- 10195224
- Publisher / Repository:
- Nature Publishing Group
- Date Published:
- Journal Name:
- Scientific Reports
- Volume:
- 10
- Issue:
- 1
- ISSN:
- 2045-2322
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
We consider testing and learning problems on causal Bayesian networks as defined by Pearl (Pearl, 2009). Given a causal Bayesian network on a graph with n discrete variables and bounded in-degree and bounded `confounded components', we show that O(logn) interventions on an unknown causal Bayesian network on the same graph, and Õ (n/ϵ2) samples per intervention, suffice to efficiently distinguish whether = or whether there exists some intervention under which and are farther than ϵ in total variation distance. We also obtain sample/time/intervention efficient algorithms for: (i) testing the identity of two unknown causal Bayesian networks on the same graph; and (ii) learning a causal Bayesian network on a given graph. Although our algorithms are non-adaptive, we show that adaptivity does not help in general: Ω(logn) interventions are necessary for testing the identity of two unknown causal Bayesian networks on the same graph, even adaptively. Our algorithms are enabled by a new subadditivity inequality for the squared Hellinger distance between two causal Bayesian networks.more » « less
-
We consider testing and learning problems on causal Bayesian networks as defined by Pearl (Pearl, 2009). Given a causal Bayesian network on a graph with n discrete variables and bounded in-degree and bounded `confounded components', we show that O(logn) interventions on an unknown causal Bayesian network on the same graph, and Õ (n/ϵ2) samples per intervention, suffice to efficiently distinguish whether = or whether there exists some intervention under which and are farther than ϵ in total variation distance. We also obtain sample/time/intervention efficient algorithms for: (i) testing the identity of two unknown causal Bayesian networks on the same graph; and (ii) learning a causal Bayesian network on a given graph. Although our algorithms are non-adaptive, we show that adaptivity does not help in general: Ω(logn) interventions are necessary for testing the identity of two unknown causal Bayesian networks on the same graph, even adaptively. Our algorithms are enabled by a new subadditivity inequality for the squared Hellinger distance between two causal Bayesian networks.more » « less
-
Probabilistic spin logic (PSL) has recently been proposed as a novel computing paradigm that leverages random thermal fluctuations of interacting bodies in a system rather than deterministic switching of binary bits. A PSL circuit is an interconnected network of thermally unstable units called probabilistic bits (p-bits), whose output randomly fluctuates between bits 0 and 1. While the fluctuations generated by p-bits are thermally driven, and therefore, inherently stochastic, the output probability is tunable with an external source. Therefore, information is encoded through probabilities of various configuration of states in the network. Recent studies have shown that these systems can efficiently solve various types of combinatorial optimization problems and Bayesian inference problems that modern computers are unfit for. Previous experimental studies have demonstrated that a single magnetic tunnel junctions (MTJ) designed to be thermally unstable can operate tunable random number generator making it an ideal hardware solution for p-bits. Most proposals for designing an MTJ to operate as a p-bit involve patterning the MTJ as a circular nano-pillar to make the device thermally unstable and then use spin transfer torque (STT) as a tuning mechanism. However, the practical realization of such devices is very challenging since the fluctuation rate of these devices are very sensitive to any device variations or defects caused during fabrication. Despite this challenge, MTJs are still the most promising hardware solution for p-bits because MTJs are very unique in that they can be tuned by multiple other mechanisms such spin orbit torque, magneto-electric coupling, and voltage-controlled exchange coupling. Furthermore, multiple forces can be used simultaneously to drive stochastic switching signals in MTJs. This means there are a large number of methods to tune, or termed as bias, MTJs that can be implemented in p-bit circuits that can alleviate the current challenges of conventional STT driven p-bits. This article serves as a review of all of the different methods that have been proposed to drive random fluctuations in MTJs to operate as a probabilistic bit. Not only will we review the single-biasing mechanisms, but we will also review all the proposed dual-biasing methods, where two independent mechanisms are employed simultaneously. These dual-biasing methods have been shown to have certain advantages such as alleviating the negative effects of device variations and some biasing combinations have a unique capability called ‘two-degrees of tunability’, which increases the information capacity in the signals generated.more » « less
-
Summary Network latent space models assume that each node is associated with an unobserved latent position in a Euclidean space, and such latent variables determine the probability of two nodes connecting with each other. In many applications, nodes in the network are often observed along with high-dimensional node variables, and these node variables provide important information for understanding the network structure. However, classical network latent space models have several limitations in incorporating node variables. In this paper, we propose a joint latent space model where we assume that the latent variables not only explain the network structure, but are also informative for the multivariate node variables. We develop a projected gradient descent algorithm that estimates the latent positions using a criterion incorporating both network structure and node variables. We establish theoretical properties of the estimators and provide insights into how incorporating high-dimensional node variables could improve the estimation accuracy of the latent positions. We demonstrate the improvement in latent variable estimation and the improvements in associated downstream tasks, such as missing value imputation for node variables, by simulation studies and an application to a Facebook data example.more » « less
