Reconfigurable Stochastic neurons based on tin oxide/MoS2 hetero-memristors for simulated annealing and the Boltzmann machine
| Autor: | Madan Dubey, Xiaodong Yan, Wei Wu, Jiang-Bin Wu, Aoyang Zhang, Jiahui Ma, Han Wang, Zhihan Zhang, Mike Shuo-Wei Chen, Matthew L. Chin, Tong Wu, Jing Guo |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: |
Multidisciplinary
ComputingMethodologies_SIMULATIONANDMODELING Computer science Science MathematicsofComputing_NUMERICALANALYSIS Physical system Statistical parameter Boltzmann machine General Physics and Astronomy General Chemistry Sigmoid function Semiconductor device Memristor ComputerSystemsOrganization_PROCESSORARCHITECTURES Nonlinear Sciences::Cellular Automata and Lattice Gases Topology Two-dimensional materials General Biochemistry Genetics and Molecular Biology Article law.invention law Simulated annealing Electronic devices Probability distribution |
| Zdroj: | Nature Communications Nature Communications, Vol 12, Iss 1, Pp 1-8 (2021) |
| ISSN: | 2041-1723 |
| Popis: | Neuromorphic hardware implementation of Boltzmann Machine using a network of stochastic neurons can allow non-deterministic polynomial-time (NP) hard combinatorial optimization problems to be efficiently solved. Efficient implementation of such Boltzmann Machine with simulated annealing desires the statistical parameters of the stochastic neurons to be dynamically tunable, however, there has been limited research on stochastic semiconductor devices with controllable statistical distributions. Here, we demonstrate a reconfigurable tin oxide (SnOx)/molybdenum disulfide (MoS2) heterogeneous memristive device that can realize tunable stochastic dynamics in its output sampling characteristics. The device can sample exponential-class sigmoidal distributions analogous to the Fermi-Dirac distribution of physical systems with quantitatively defined tunable “temperature” effect. A BM composed of these tunable stochastic neuron devices, which can enable simulated annealing with designed “cooling” strategies, is conducted to solve the MAX-SAT, a representative in NP-hard combinatorial optimization problems. Quantitative insights into the effect of different “cooling” strategies on improving the BM optimization process efficiency are also provided. Boltzmann Machines offer the potential of more efficient solutions to combinatorial problems compared to von Neumann computing architectures. Here, Yan et al introduce a stochastic memristor with dynamically tunable properties, a vital feature for the efficient implementation of a Boltzmann Machine. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full text from SpringerLink