Photonic Neural Networks with Spatiotemporal Dynamics : : Paradigms of Computing and Implementation.

• Author: Suzuki, Hideyuki.
• TeilnehmendeR: Tanida, Jun.; Hashimoto, Masanori.
• Place / Publishing House: Singapore : : Springer,, 2023.; ©2024.
• Year of Publication: 2023
• Edition: 1st ed.
• Language: English
• Physical Description: 1 online resource (277 pages)

Table of Contents

• Intro
• Preface
• Contents
• Introduction
• Revival of Optical Computing
• 1 Introduction
• 2 Fourier Optical Processing
• 3 Digital Optical Computing
• 4 Optical Neurocomputing
• 5 Optical Special Purpose Processing
• 6 Nanophotonic Computing
• 7 Toward Further Extension
• References
• Nonlinear Dynamics and Computing in Recurrent Neural Networks
• 1 Introduction
• 2 Fundamental RNN Models and Energy Function
• 2.1 Amari-Hopfield Network with Binary States
• 2.2 Amari-Hopfield Network with Continuous States
• 2.3 Boltzmann Machine
• 3 Nonlinear Dynamics in Symmetrically Connected RNNs
• 3.1 Chaotic Neural Network
• 3.2 Chaotic Boltzmann Machine
• 3.3 Ising Machines
• 3.4 Herding System
• 4 Reservoir Computing
• 4.1 Training Input-Output Relation of RNNs
• 4.2 Echo State Network
• 4.3 Echo State Property and Reservoir Design
• 4.4 Neural Network Reservoirs
• 4.5 Physical Reservoir Computing
• 5 Towards Photonic Neural Network Computing
• References
• Fluorescence Energy Transfer Computing
• Fluorescence Energy Transfer Computing
• 1 Fluorescence Energy Transfer
• 2 FRET-Based Device
• 2.1 DNA for FRET Devices
• 2.2 FRET-Based Photonic Wire
• 2.3 FRET-Based Photonic Switch
• 2.4 FRET-Based Logic Gate
• 3 Scaffold DNA Logic
• 3.1 Basic Concept
• 3.2 Arrangement Control of Fluorescent Molecules
• 3.3 Design of Logic Operations
• 3.4 AND Operation
• 3.5 OR Operation
• 3.6 NOT Operation
• 3.7 Extended FRET Connection
• 4 Optical Program of DNA Scaffold Logic
• 4.1 Optical Control of FRET
• 4.2 Optical Control of FRET Pathway
• 4.3 Optically Programmable DNA Scaffold Logic
• 5 FRET Network-Based Computation
• References
• Quantum-Dot-Based Photonic Reservoir Computing
• 1 Introduction
• 2 Quantum-Dot-Based Physical Reservoir
• 2.1 Basics
• 2.2 Experimental Demonstration: Randomly-Dispersed QDR.
• 2.3 Experimental Demonstration: Electrophoresis
• 3 Non-linearity of the QD Reservoir
• 3.1 Experimental Setup
• 3.2 Qualitative Non-linearity
• 4 Spatio-Temporal Photonic Processing
• 4.1 Basics
• 4.2 Streak Measurement
• 4.3 Spatiotemporal Reservoir Model
• 4.4 Experimental Demonstration
• 5 Conclusion and Future Prospect
• References
• Exploring Integrated Device Implementation for FRET-Based Optical Reservoir Computing
• 1 Introduction
• 2 Proposed Device Structure
• 2.1 Device Structure
• 2.2 Network Mapping
• 2.3 Experiments
• 3 Proof-of-Concept Prototyping
• 3.1 Implementation
• 3.2 Evaluation
• 4 Discussion on Energy Advantage
• 4.1 Power Estimation Approach
• 4.2 Result
• 5 Summary
• References
• FRET Networks: Modeling and Analysis for Computing
• 1 Introduction
• 2 Spatiotemporal Model for FRET Networks
• 2.1 Multiple-Donor Model
• 2.2 Analytical Results
• 3 Simulation Methods
• 3.1 Deterministic Simulation
• 3.2 Stochastic Simulation
• 3.3 Comparison Between Deterministic and Stochastic Simulation
• 4 Information Processing Using FRET Networks
• 4.1 Nonlinear Function Approximation
• 4.2 Chaotic Time-Series Prediction
• 4.3 Handwritten Digit Recognition
• 5 Conclusions and Future Works
• References
• Quantum Walk on FRET Networks
• 1 Introduction of Quantum Walk
• 1.1 Classical Random Walk
• 1.2 Quantum Walk
• 1.3 Quantum Walk on FRET Networks
• 2 Application of Quantum Walk
• 2.1 Classical Parton Shower
• 2.2 Quantum Parton Shower Algorithm
• 3 Conclusion
• References
• Spatial-Photonic Spin System
• Spatial Photonic Ising Machine with Time/Space Division Multiplexing
• 1 Introduction
• 2 Spatial Photonic Ising Machine with Multiplexing
• 2.1 Basic Scheme of SPIM
• 2.2 Concept of SPIMs with Multiplexing
• 3 Time Division Multiplexed (TDM)-SPIM.
• 4 Space Division Multiplexing (SDM)-SPIM
• 5 Conclusion
• References
• Investigation on Oscillator-Based Ising Machines
• 1 Introduction
• 2 Ising Model and Ising Machines
• 3 Oscillator-Based Ising Machines
• 3.1 Phase Oscillators
• 3.2 Second-Harmonic Injection Locking
• 3.3 Kuramoto Model
• 3.4 Gradient Structure of the Kuramoto Model
• 3.5 Working Principle of OIMs
• 4 Experiments
• 4.1 MAX-CUT Problem
• 4.2 Experimental Setting and Evaluation Metrics
• 4.3 Effect of Properties of Noisy Fluctuation and Coarse Time Discretization
• 4.4 Effect of Injection and Noise Strength
• 4.5 Effect of Higher Harmonics in Coupling and Injection Schemes
• 5 Summary
• References
• Sampling-Like Dynamics of the Nonlinear Dynamical System Combined with Optimization
• 1 Herding: Sample Generation Using Nonlinear Dynamics and Monte Carlo Integration
• 1.1 Deterministic Tossup
• 1.2 Herding
• 1.3 Convergence of Herding
• 2 Herded Gibbs: Model-Based Herding on Spin System
• 2.1 Markov Chain Monte Carlo Method
• 2.2 Herded Gibbs
• 2.3 Sharing Weight Variables
• 2.4 Monte Carlo Integration Using HG
• 2.5 Rao-Blackwellization
• 3 Entropic Herding: Regularizer for Dynamical Sampling
• 3.1 Maximum Entropy Principle
• 3.2 Entropic Herding
• 3.3 Dynamical Entropic Maximization in Herding
• 3.4 Other Notes on Entropic Herding
• 4 Discussion
• 4.1 Application of Physical Combinatorial Optimization Using Herding
• 4.2 Generative Models and Herding
• 5 Conclusion
• References
• Photonic Reservoir Computing
• Reservoir Computing Based on Iterative Function Systems
• 1 Introduction
• 2 Iterative Function Systems
• 3 Iterative Function System-Based Reservoir Computing
• 4 Prediction Performance of IFSRC
• 4.1 Multi-step Ahead Prediction
• 4.2 Single-Step Ahead Prediction of Santa Fe Time-Series Data.
• 5 Experimental Performance of IFS Reservoir Computing
• 5.1 Optical Setup
• 5.2 Multi-step Ahead Prediction of Mackey-Glass Equation
• 5.3 Single-step Ahead Prediction of Santa Fe Time-Series Data
• 6 Relationship Between Performance and Spectral Radius
• 7 IFS Reservoir Computing with Optical Feedback Loop
• 8 Discussion
• References
• Bridging the Gap Between Reservoirs and Neural Networks
• 1 Introduction
• 2 Background
• 2.1 Lottery Ticket Hypotheses
• 2.2 Residual Neural Networks
• 3 Hidden-Fold Networks
• 3.1 Folded ResNet Architecture
• 3.2 Supermask Training
• 4 Experiments and Results
• 4.1 Experimental Settings
• 4.2 Results
• 5 Summary
• References
• Brain-Inspired Reservoir Computing Models
• 1 Introduction
• 2 Reservoir-Based Predictive Coding Model
• 3 Reservoir-Based Reinforcement Learning Model
• 4 Integrated Model and Mental simulation
• 5 Summary
• References.