A Study of Gaussian Activation Function Based Modular Neural Network for Alternative-Style Handwritten Characters Recognition System
スポンサーリンク
概要
- 論文の詳細を見る
We propose a design method using Gaussian activation function for alternative-style handwritten character recognition system. While the alternative method can gain high accuracy recognition performance by simplifying the recognition problem to linear discriminant problem, the overfitting problem will occur with small number of learning samples. In this paper, we introduce Gaussian function as activation function of neurons in order to avoid the overfitting problem. Our proposed method can learn the overall distribution of samples and gain higher generalization ability. In the recognition experiment using ETL9B 3036 categories, the proposed method can achieve 97.67% recognition accuracy.
- 東北大学の論文
著者
-
KATO Nei
Graduate School of Information Sciences, Tohoku University
-
NEMOTO Yoshiaki
Graduate School of Information Sciences, Tohoku University
-
Ishii Tsutomu
Matsushita Communication Industrial Co. Ltd. Software Development Center
-
Nemoto Y
Graduate School Of Information Sciences (gsis) Tohoku University
-
Kato Nei
Graduate School Of Information Sciences (gsis) Tohoku University
-
Nemoto Yoshiaki
Graduate School Of Information Science Tohoku University
-
WAIZUMI Yuji
Graduate School of Information Sciences, Tohoku University
-
Waizumi Yuji
Graduate School Of Information Sciences (gsis) Tohoku University
関連論文
- On the effect of mobility on reliable topology control in ad-hoc networks (ネットワークシステム)
- On the Effect of Mobility on Reliable Topology Control in Ad-hoc Networks
- Characterization of Packet Losses and Delays in Multi-Layered Satellite Networks Over an Efficient Routing Scheme
- A secure and robust framework of service level specification for next generation wireless networks (ユビキタス・センサネットワーク)
- Problems in Management Information Retrieval for High-Speed Networks and a Proposed Solution (Special Issue on Multimedia on Demand)
- B-11-10 A Study on QoS guarantee for Multiclass Traffic
- Hand-written Character Recognition System Using Uniform Division of Training Samples on subspace Method
- Performance evaluation of TCP over hybrid satellite/MANET networks (衛星通信)
- Capacity Design of Guaranteed-QoS VPN
- A Satellite Selection Method for Walker Delta LEO Satellite Networks(Recent Fundamental Technologies for Broadband Satellite Communications)
- A New Coding Scheme for Infrared ASK Communication Systems Which Have Zero-Spectrum Points within the Main-Lobe Region (Joint Special Issue on Recent Progress in Optoelectronics and Communications)
- A New Coding Scheme for Infrared ASK Communication Systems Which Have Zero-Spectrum Points within the Main-Lobe Region (Joint Special Issue on Recent Progress in Optoelectronics and Communications)
- A Fundamental Study on Effect of Contact Condition for Electromagnetic Noise at Copper-Carbon Electrodes (Special Issue on Electromechanical Devices and Their Materials)
- Usage Parameter Control Scheme for ATM Networks : Markovian State-Dependent Approach
- High Accuracy Recognition of ETL9B Using Exclusive Learning Neural Network-II : ELNET-II (Special Issue on Character Recognition and Document Understanding)
- ELB in IP/NGEO Satellite Constellations : Issues and Solutions(Advanced Technology-I)
- A Study of Gaussian Activation Function Based Modular Neural Network for Alternative-Style Handwritten Characters Recognition System
- Characterization of Packet Losses and Delays in Multi-Layered Satellite Networks Over an Efficient Routing Scheme
- Detecting and Tracing DDoS Attacks in the Traffic Analysis Using Auto Regressive Model(Traffic Measurement and Analysis)(New Thechnologies and their Applications of the Internet)
- High Speed and High Accuracy Rough Classification for Handwritten Characters Using Hierarchical Learning Vector Quantization
- A Handprinted Character Recognition System Using Image Transformation Based on Partial Inclination Detection (Special Issue on Character Recognition and Document Understanding)
- A Traffic Congestion Avoidance Scheme for LEO Satellite Networks by Using Traffic Detour Technique(System Engineering-II)
- B-6-128 Admission Control and Resource Allocation for Hierarchical-Generalized Processor Sharing System
- Congestion Prediction Based Load Balancing in IP/LEO Satellite Networks(Satellite Network)
- Network Application Identification Using Sequential Transition Patterns of Payload Length