Novel hardware and concepts for unconventional computing

Ziegler, Martin

doi:10.1038/s41598-020-68834-1

Cited by 15 publications

(11 citation statements)

References 27 publications

(27 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Promising attempts to the online embodied evolution of robots have been proposed [ 67 , 68 ]. Soft robotics [ 69 ], and unconventional computing systems [ 70 ] may provide a viable approach to the evolution of sensors and actuators, along with self-improvement of behavior policies (which, of course, may greatly benefit from current machine learning and AI techniques).…”

Section: Discussion and Conclusionmentioning

confidence: 99%

Emergence of Organisms

Roli

Kauffman

2020

Entropy

View full text Add to dashboard Cite

Since early cybernetics studies by Wiener, Pask, and Ashby, the properties of living systems are subject to deep investigations. The goals of this endeavour are both understanding and building: abstract models and general principles are sought for describing organisms, their dynamics and their ability to produce adaptive behavior. This research has achieved prominent results in fields such as artificial intelligence and artificial life. For example, today we have robots capable of exploring hostile environments with high level of self-sufficiency, planning capabilities and able to learn. Nevertheless, the discrepancy between the emergence and evolution of life and artificial systems is still huge. In this paper, we identify the fundamental elements that characterize the evolution of the biosphere and open-ended evolution, and we illustrate their implications for the evolution of artificial systems. Subsequently, we discuss the most relevant issues and questions that this viewpoint poses both for biological and artificial systems.

show abstract

Section: Discussion and Conclusionmentioning

confidence: 99%

Emergence of Organisms

Roli

Kauffman

2020

Entropy

View full text Add to dashboard Cite

show abstract

“…As particle-like highly localized objects, solitons can carry and exchange information, which make them unique entities for unconventional computation [1][2][3][4][5][6][7] . Robustness to perturbations and very importantly to collisions is an essential ingredient to build soliton-based nanoelectronics.…”

Section: Introductionmentioning

confidence: 99%

Tailoring Elastic Scattering of Relativistic Antiferromagnetic Domain Walls for Collision-based Computing

Otxoa

Tatara

Roy

et al. 2023

Preprint

View full text Add to dashboard Cite

Soliton-based computing relies on their unique properties for transporting energy and emerging intact from head-on collisions. Magnetic domain walls are often referred to as solitons disregarding the strict mathematical definition requiring the above scattering property. Here we demonstrate the conditions of elastic and inelastic scattering for spinorbit torque-induced dynamics of relativistic domain walls on the technologically relevant Mn2Au antiferromagnetic material. We show that even domain walls with opposite winding numbers can experience elastic scattering and we present the corresponding phase diagram as a function of the spin-orbit field strength and duration. The elastic collision requires minimum domain walls speed, which we explain assuming an attractive potential created by domain wall pair. On the contrary, when the domain walls move at lower speeds, their collision is inelastic and results in a dispersing breather. Our findings will be important for the development of soliton-based computing using antiferromagnetic spintronics and we discuss their prospects for building NOT and XOR gates.

show abstract

“…On the flipside of this revolution is the fact that training neural networks is a computationally expensive task that has to be performed on resource-intensive high-performance computing hardware. This is starting to raise serious concerns about economical and ecological sustainability 1 , 2 , which has instigated an intensive search for alternative computing systems, such as quantum annealers or hybrid analog-digital computing concepts 3 – 6 , that can perform training of neural networks significantly faster and more efficiently than current generations of digital computers 7 , 8 . Among this drive for more efficient computing concepts, analog Ising machines have emerged as a promising solution 9 , 10 .…”

Section: Introductionmentioning

confidence: 99%

Noise-injected analog Ising machines enable ultrafast statistical sampling and machine learning

et al. 2022

View full text Add to dashboard Cite

Ising machines are a promising non-von-Neumann computational concept for neural network training and combinatorial optimization. However, while various neural networks can be implemented with Ising machines, their inability to perform fast statistical sampling makes them inefficient for training neural networks compared to digital computers. Here, we introduce a universal concept to achieve ultrafast statistical sampling with analog Ising machines by injecting noise. With an opto-electronic Ising machine, we experimentally demonstrate that this can be used for accurate sampling of Boltzmann distributions and for unsupervised training of neural networks, with equal accuracy as software-based training. Through simulations, we find that Ising machines can perform statistical sampling orders-of-magnitudes faster than software-based methods. This enables the use of Ising machines beyond combinatorial optimization and makes them into efficient tools for machine learning and other applications.

show abstract

Novel hardware and concepts for unconventional computing

Cited by 15 publications

References 27 publications

Emergence of Organisms

Emergence of Organisms

Tailoring Elastic Scattering of Relativistic Antiferromagnetic Domain Walls for Collision-based Computing

Noise-injected analog Ising machines enable ultrafast statistical sampling and machine learning

Contact Info

Product

Resources

About