Avoidance of the Permanent Oscillating State in the Inverse Function Delayed Neural Network(Neuron and Neural Networks,<Special Section>Nonlinear Theory and its Applications)
スポンサーリンク
概要
- 論文の詳細を見る
Many researchers have attempted to solve the combinatorial optimization problems, that are NP-hard or NP-complete problems, by using neural networks. Though the method used in a neural network has some advantages, the local minimum problem is not solved yet. It has been shown that the Inverse Function Delayed (ID) model, which is a neuron model with a negative resistance on its dynamics and can destabilize an intended region, can be used as the powerful tool to avoid the local minima. In our previous paper, we have shown that the ID network can separate local minimum states from global minimum states in case that the energy function of the embed problem is zero. It can achieve 100% success rate in the N-Queen problem with the certain parameter region. However, for a wider parameter region, the ID network cannot reach a global minimum state while all of local minimum states are unstable. In this paper, we show that the ID network falls into a particular permanent oscillating state in this situation. Several neurons in the network keep spiking in the particular permanent oscillating state, and hence the state transition never proceed for global minima. However, we can also clarify that the oscillating state is controlled by the parameter α which affects the negative resistance region and the hysteresis property of the ID model. In consequence, there is a parameter region where combinatorial optimization problems are solved at the 100% success rate.
- 社団法人電子情報通信学会の論文
- 2007-10-01
著者
-
Hayakawa Yoshihiro
Laboratory For Brainware Reseach Institute Of Electrical Comunication Tohoku University:laboratory F
-
Nakajima Koji
Laboratory For Brainware Reseach Institute Of Electrical Comunication Tohoku University:laboratory F
-
SATO Akari
Laboratory for Brainware Reseach Institute of Electrical Comunication, Tohoku University
-
Sato Akari
Laboratory For Brainware Reseach Institute Of Electrical Comunication Tohoku University:laboratory F
関連論文
- Retrieval Property of Associative Memory Based on Inverse Function Delayed Neural Networks(Nonlinear Problems)
- Design of a Neural Network Chip for the Burst ID Model with Ability of Burst Firing(Selected Papers from the 19th Workshop on Circuits and Systems in Karuizawa)
- Dynamical Behavior of Neural Networks with Anti-Symmetrical Cyclic Connections(Oscillation, Dynamics and Chaos,Nonlinear Theory and its Applications)
- Recalling Temporal Sequences of Patterns Using Neurons with Hysteretic Property
- Temporal Sequences of Patterns with an Inverse Function Delayed Neural Network(Control, Neural Networks and Learning,Nonlinear Theory and its Applications)
- New Nonvolatile Analog Memories for Analog Data Processing
- New Nonvolatile Analog Memories for Building Associative Memories
- Toward an Ideal Synthesis of Oligonucleotides : Development of a Novel Phosphoramidite Method with High Capability^#
- Magnetic Isolation on a Superconducting Ground Plane
- Numerical Investigation and Model Approximation for the Hysteretic Current-Voltage Characteristics of Josephson Junctions with Nonlinear Quasiparticle Resistance
- Single Electron Stochastic Neural Network(Nonlinear Theory and its Applications)
- Single Electron Random Number Generator(Electronic Circuits)
- Hardware Implementation of New Analog Memory for Neural Networks
- LSI Neural Chip of Pulse-Output Network with Programmable Synapse
- A Content-Addressable Memory Using "Switched Diffusion Analog Memory with Feedback Circuit" (Special Section on Analog Circuit Techniques and Related Topics)
- Switched Diffusion Analog Memory for Neural Networks with Hebbian Learning Function and Its Linear Operation (Special Section of Papers Selected from JTC-CSCC'95)
- Limit Cycles of One-Dimensional Neural Networks with the Cyclic Connection Matrix (Special Section of Papers Selected from JTC-CSCC'95)
- Superconducting neural circuits using stochastic logic and new fabrication process elements
- High Throughput Parallel Arithmetic Circuits for Fast Fourier Transform
- Comparison between an AND Array and a Booth Encoder for Large-Scale Phase-Mode Multipliers(Special Issue on Superconductor Digital/Analog Circuit Technologies)
- Implementation of Continuous-Time Dynamics on Stochastic Neurochip(Nonlinear Theory and its Applications)
- Integrated Circuits of Map Chaos Generators (Special Section on Analog Circuit Techniques and Related Topics)
- Analog CMOS Implementation of Quantized Interconnection Neural Networks for Memorizing Limit Cycles (Special Section of Papers Selected from ITC-CSCC '98)
- Digital Circuits Based on Single Flux Quanta
- Avoidance of the Permanent Oscillating State in the Inverse Function Delayed Neural Network(Neuron and Neural Networks,Nonlinear Theory and its Applications)
- Hardware Neural Network for a Visual Inspection System
- Hardware Implementation of a DBM Network with Non-monotonic Neurons
- Analysis of burst dynamics bound by potential with active areas
- Early activation and interferon-γ production of tumor-infiltrating mature CD27^ natural killer cells
- Quantum Neural Network Composed of Kane's Qubits
- An Approach for Quantum Computing using Adiabatic Evolution Algorithm