Neural Information Processing

Name: Neural Information Processing | 25th International Conference, ICONIP 2018, Siem Reap, Cambodia, December 13-16, 2018, Proceedings, Part I
Brand: Springer
Price: 53.49 EUR
Availability: OnlineOnly

25th International Conference, ICONIP 2018, Siem Reap, Cambodia, December 13-16, 2018, Proceedings, Part I

Long Cheng Andrew Chi Sing Leung Seiichi Ozawa(Editor)

Springer (Publisher)

Published on 3. December 2018

XXII, 647 pages

E-Book

PDF with digital watermarking

System requirements

978-3-030-04167-0 (ISBN)

€53.49incl. 7% vat

System requirements

for PDF with digital watermarking

E-Book Single Licence

Available for download

Description

More details

Other editions

Content

Intro
Preface
ICONIP 2018 Organization
Contents - Part I
Deep Neural Networks
Adaptive Deep Dictionary Learning for MRI Reconstruction
Abstract
1 Introduction
2 Literature Review
3 Proposed Formulation
4 Experimental Evaluation
4.1 Dataset Description
4.2 Results
5 Conclusion
Acknowledgements
References
Deep-PUMR: Deep Positive and Unlabeled Learning with Manifold Regularization
1 Introduction
2 Preliminaries
2.1 Non-negative PU Learning
2.2 Manifold Regularization
3 The Proposed Deep-PUMR Model
3.1 Network Architecture
3.2 Loss Function
4 Experiments
5 Conclusion
References
Viewpoint Estimation for Workpieces with Deep Transfer Learning from Cold to Hot
1 Introduction
2 Related Work
3 Viewpoint Estimation for Workpieces
3.1 Deep Transfer Network and Loss Function
3.2 Cold-to-Hot Training
4 Experimental Results
4.1 Experiment Setup
4.2 Viewpoint Estimation Results
4.3 Visualization of Learned Transfer Features
5 Conclusion
References
Co-consistent Regularization with Discriminative Feature for Zero-Shot Learning
1 Introduction
2 Related Work
3 Proposed Method
3.1 Iteration Network
3.2 Visual-Semantic Cohesion Network
3.3 Co-consistent Regularization
4 Experiment
4.1 Experimental Setup
4.2 Experimental Results
5 Conclusions
References
Hybrid Networks: Improving Deep Learning Networks via Integrating Two Views of Images
1 Introduction
2 Background
2.1 PCANet
2.2 Tensor Preliminaries
3 The Tensor Factorization Network (TFNet)
3.1 The First Layer
3.2 The Second Layer
4 The Hybrid Network (HybridNet)
4.1 The First Layer
4.2 The Second Layer
5 Experiments and Results
5.1 Experimental Setup
5.2 Datasets
5.3 Results and Discussions
6 Conclusion and Future Work
References
On a Fitting of a Heaviside Function by Deep ReLU Neural Networks
1 Introduction
2 Approximation of a Heaviside Function
2.1 Properties of ReLU
2.2 Approximation by a Simple Network
2.3 Approximation by a Deep Structure
3 Training Issue
3.1 Effect of Deep Structure in Learning
3.2 A Simple Numerical Example
4 Conclusions and Future Works
References
Deep Tag Recommendation Based on Discrete Tensor Factorization
1 Introduction
2 Preliminaries and Problem Statement
2.1 Notations
2.2 Problem Statement
3 Discrete Tensor Factorization Model
3.1 Content Enhanced Model
4 Experiments
4.1 Experimental Settings
4.2 Efficiency Comparison (RQ1)
4.3 Accuracy Comparison (RQ2)
5 Conclusion
References
Efficient Integer Vector Homomorphic Encryption Using Deep Learning for Neural Networks
1 Introduction
2 Preliminaries
2.1 Fully Homomorphic Encryption
2.2 Concept of Deep Learning
3 Efficient Integer Vector Homomorphic Encryption Scheme for Neural Networks
3.1 Efficient Integer Vector Homomorphic Encryption (EIVHE)
3.2 Neural Networks
4 Experiments
4.1 Dataset
4.2 Baseline Model
4.3 EIVHE Model
5 Performance Evaluation
5.1 Accuracy on the Separate Datasets
5.2 Accuracy on the Improved Algorithm
5.3 Accuracy on the Absolute Datasets
6 Conclusions
References
DeepSIC: Deep Semantic Image Compression
1 Introduction
2 Proposed Deep Semantic Image Compression System
2.1 Feature Extraction
2.2 Binarizer
2.3 Reconstruction from Features
2.4 Semantic Analysis
2.5 Joint Training of Compression and Semantic Analysis
3 Experiment
3.1 Experimental Setup
3.2 Experimental Results
4 Conclusion and Future Work
References
Multi-stage Gradient Compression: Overcoming the Communication Bottleneck in Distributed Deep Learning
1 Introduction
2 Related Works
2.1 Stages in Machine Learning Training
2.2 Solutions to Communication Bottleneck
3 Multi-stage Gradient Compression
3.1 Motivation
3.2 Multi-stage Gradient Compression
3.3 Merged Residual Correction
4 Experiments
4.1 Experiment Settings
4.2 Convergence Research of MGC
4.3 Gradient Compression Ratio and Accuracy on ImageNet
4.4 Scalability and Speedup Performance
5 Conclusion
References
Teach to Hash: A Deep Supervised Hashing Framework with Data Selection
1 Introduction
2 Methodology
2.1 Deep Hashing Network
2.2 Teach to Hash
3 Experiment
3.1 Datasets and Settings
3.2 Effect of Teacher
3.3 Results on CIFAR-10 and NUS-WIDE
3.4 Parameter Sensitivity
4 Conclusion
References
Multi-view Deep Gaussian Processes
1 Introduction
2 Deep Gaussian Processes
3 Multi-view Deep Gaussian Processes
3.1 The Proposed Model
3.2 Variational Bayesian Training
3.3 Parameter Estimation
4 Experiments
4.1 Data Sets
4.2 Setting
4.3 Results
5 Conclusion
References
Deep Collaborative Filtering Combined with High-Level Feature Generation on Latent Factor Model
1 Introduction
2 Preliminaries
2.1 Implicit Feedback
2.2 Latent Factor Model
2.3 Neural Networks
3 Deep Collaborative Filtering
3.1 Feature Mapping
3.2 Weighted Feature Interaction Network
3.3 User-Item Interaction Modeling
4 Experiments
4.1 Experimental Settings
4.2 Performance
5 Related Work
6 Conclusions
References
Data Imputation of Wind Turbine Using Generative Adversarial Nets with Deep Learning Models
Abstract
1 Introduction
2 GAN Based Data Interpolation Method
2.1 Generative Adversarial Nets
2.2 Interpolation Method
3 Case Study
3.1 Experiment Settings
3.2 Comparative Experiments
4 Conclusion
References
A Deep Ensemble Network for Compressed Sensing MRI
1 Introduction
2 Related Works
2.1 Sparse-Optimization Methods
2.2 Deep Neural Network Models
2.3 Ensemble Learning in Deep Learning
3 Method
3.1 Intra-block Ensemble
3.2 Inter-block Ensemble
4 Experiments
4.1 Experimental Setup
4.2 Results
4.3 Discussion on Hyperparameter
4.4 Experiments on Another Dataset
4.5 Running Time
5 Conclusion
References
Deep Imitation Learning: The Impact of Depth on Policy Performance
Abstract
1 Introduction
2 Convolutional Neural Network
3 Experimental Setup
4 Results and Discussions
5 Future Works
6 Conclusion
References
Improving Deep Neural Network Performance with Kernelized Min-Max Objective
1 Introduction
2 Proposed Method
2.1 General Framework
2.2 Min-Max Strategy
2.3 Kernelized Min-Max Objective
3 Experiments
3.1 Setting
3.2 On Shallow CNN
3.3 On Deep CNN
3.4 On Deep Residual Model
4 Conclusion
References
Understanding Deep Neural Network by Filter Sensitive Area Generation Network
Abstract
1 Introduction
2 Filter Sensitive Area Generation Network
2.1 Filters Selection
2.2 Filter Sensitive Area Generation Network
3 Experiments
3.1 Implementation Details
3.2 Experiments on MNIST
3.3 Experiments on FGVC-Aircraft
4 Discussion
5 Conclusion and Future Work
Acknowledgements
References
Accelerating Deep Q Network by Weighting Experiences
1 Introduction
2 Deep Q Network
3 Proposed Method
4 Experiment and Results
4.1 Environment
4.2 Settings
4.3 Results
5 Related Works
6 Conclusion
References
Exploring Deep Learning Architectures Coupled with CRF Based Prediction for Slot-Filling
1 Introduction
2 Related Works
2.1 Background
2.2 Motivation and Contribution
3 Proposed Methodology
3.1 Baseline Models
3.2 Proposed Models
4 Experimentation, Results and Analysis
4.1 Experimentation
4.2 Results and Analysis
4.3 Error Analysis
5 Conclusions and Future Work
References
Domain Adaptation via Identical Distribution Across Models and Tasks
1 Introduction
2 Related Work
3 Transfer CNN Distribution Knowledge Across Model and Domain
4 Evaluation
4.1 Application on Handwritten Digit Dataset
4.2 Distribution Loss Applies on Disparate Domains
4.3 Adaptation on the Office Dataset
5 Discussion and Conclusion
References
A Pointer Network Based Deep Learning Algorithm for the Max-Cut Problem
1 Introduction
2 Problem Formulation
2.1 The Max-Cut Problem
2.2 The Pointer Network
3 Solving the Max-Cut Problem Using the Pointer Network
3.1 Data Structure of the Max-Cut Problem
3.2 Datasets Generation
3.3 Supervised Learning
4 Experiments Results and Analysis
5 Conclusion
References
Convolution Neural Networks
Multi-stream with Deep Convolutional Neural Networks for Human Action Recognition in Videos
Abstract
1 Introduction
2 Related Work
3 Network Architecture
3.1 3D CNNs
3.2 CNNs and Attention Block
3.3 Multi-stream
4 Experiments
4.1 Datasets
4.2 Implementations Details
4.3 Evaluation
5 Conclusion
References
Use 3D Convolutional Neural Network to Inspect Solder Ball Defects
Abstract
1 Introduction
2 Related Works
3 Approach
3.1 3D Solder Balls Volumetric Data
3.2 3D Convolutional Neural Network
4 Experiments
4.1 Solder Ball Data
4.2 Preprocessing Acceleration Comparison
4.3 Experimental Result
5 Conclusions and Future Work
Acknowledgement
References
User-Invariant Facial Animation with Convolutional Neural Network
Abstract
1 Introduction
2 CNN for Face Animation
2.1 Data Preparation
2.2 Network Architecture
2.3 Loss Function
3 Experiments
4 Conclusion
Acknowledgments
References
An Approach for Feature Extraction and Diagnosis of Motor Rotor Bearing Based on Convolution Neural Network
Abstract
1 Introduction
2 Phase Space Reconstruction and Data Processing
2.1 Selection of Time Delay and Embedding Dimension
2.2 Data Processing and Grayscale Map
3 Convolutional Neural Network (CNN)
4 Simulation Results and Analysis
4.1 Sample Set Preprocessing
4.2 Selection of Time Delay and Time Window
4.3 Composition Reconstruction Grayscale Image
4.4 Result of Fault Diagnosis
5 Conclusion
References
Gated Convolutional Networks for Commonsense Machine Comprehension
1 Introduction
2 Task Description
3 Model
3.1 Input Layer
3.2 Gated Convolutional Layer
3.3 Output Layer
4 Experiments and Discussion
4.1 Setup
4.2 Results
4.3 Ablation Study
4.4 Self-Attention vs Max Pooling
4.5 Gating Mechanisms
5 Conclusion
References
Cursive Scene Text Analysis by Deep Convolutional Linear Pyramids
1 Introduction
2 Related Work
3 Proposed Methodology
3.1 Formulation of Linear Spatial Pyramids of Cursive Arabic Scene Text
3.2 Preprocessing of Image Pyramids by Image Filters
3.3 LSTM and It's Implicit Segmentation
4 Experimental Analysis
4.1 Scene Text Dataset
5 Conclusion
References
Human Action Recognition with 3D Convolution Skip-Connections and RNNs
1 Introduction
2 Related Work
3 Network Architecture
3.1 3D Convolution Skip-Connections
3.2 The Batch Normalization Layer
3.3 The RNN
4 Experiments
4.1 Dataset and Experimental Protocols
4.2 Comparison of Different Architectures Variants
4.3 The Influence of the Batch Normalization Layer
4.4 Comparison with Other Excellent Methods
5 Conclusions
References
Convolutional Neural Network with Discriminant Criterion for Input of Each Neuron in Output Layer
1 Introduction
2 CNN with Discriminant Criterion
3 Experiments
3.1 Datasets
3.2 Network Architecture
3.3 Classification Performance
3.4 Distribution of Input Values
3.5 Visualization of Input Values
4 Conclusion
References
Proposal of Complex-Valued Convolutional Neural Networks for Similar Land-Shape Discovery in Interferometric Synthetic Aperture Radar
1 Introduction
2 Complex-Valued Convolutional Neural Networks
2.1 Construction of Complex-Valued Convolutional Neural Networks
2.2 Complex-Valued Convolutional Processing
2.3 Learning Dynamics
2.4 Complex-Valued Pooling Process
2.5 Decision Network Having Complex-Valued Fully Connected Neurons
3 Experiments
3.1 Preprocessing of InSAR Images
3.2 Parameters in the Neural Network
3.3 Teacher Images and the Learning Process
4 Results
5 Conclusion
References
Feature Learning and Transfer Performance Prediction for Video Reinforcement Learning Tasks via a Siamese Convolutional Neural Network
1 Introduction
2 Preliminary
2.1 Reinforcement Learning Tasks
2.2 Transfer Learning Scenario
3 Predicting Transfer Performance
3.1 Problem Formulation and Training Data
3.2 Predicting Transfer Performance via Deep Neural Network
4 Experimental Evaluation
4.1 Maze Domain
4.2 Ms. PacMan Domain
5 Conclusion
References
Structured Sequence Modeling with Graph Convolutional Recurrent Networks
1 Introduction
2 Preliminaries
2.1 Structured Sequence Modeling
2.2 Convolutional Neural Networks on Graphs
3 Related Works
4 Proposed GCRN Models
5 Experiments
5.1 Spatio-Temporal Sequence Modeling on Moving-MNIST
5.2 Natural Language Modeling on Penn Treebank
6 Conclusion and Future Work
References
Part-Level Sketch Segmentation and Labeling Using Dual-CNN
1 Introduction
2 Related Works
3 Proposed Method
3.1 Modelling Postion and Orientation with Triple Channels
3.2 Dual Convolutional Neural Networks
4 Experiments and Results
4.1 Dataset and Setting
4.2 Effect of Individual Components
4.3 Comparison with State of the Art and CNN Baseline
4.4 Runtime Efficiency
5 Discussion
6 Conclusion
References
RE-CNN: A Robust Convolutional Neural Networks for Image Recognition
1 Introduction
2 Related Work
3 The Proposed RE-CNN Model
4 Experiments
4.1 Implementation Details
4.2 Discussion About the Split Parameter k
4.3 Results
5 Conclusion
References
MusicCNNs: A New Benchmark on Content-Based Music Recommendation
1 Introduction
2 Related Work
3 The Collected Data Set
4 Music Convolution Neural Networks (MusicCNNs)
4.1 The Architecture of MusicCNNs
4.2 Regularization Methods
5 Experiments
5.1 Quantitative Evaluation
5.2 Qualitative Evaluation
6 Conclusion
References
Remote Sensing Image Segmentation by Combining Feature Enhanced with Fully Convolutional Network
1 Introduction
2 Related Work
3 Main Idea and Methods
3.1 Data Pretreatment
3.2 FCN
4 Experiment
4.1 Data
4.2 Experiment Result and Comparison
5 Conclusion and Future Work
References
A New LSTM Network Model Combining TextCNN
1 Introduction
1.1 Task Definition
2 Background
2.1 Word Vectors
2.2 TextCNN
2.3 Long Short Term Memory
3 TC-LSTM
3.1 Feature Combination
4 Experiments and Analysis
4.1 Experiment Results
5 Conclusion
References
Self-inhibition Residual Convolutional Networks for Chinese Sentence Classification
1 Introduction
2 SIRCNN Architecture
2.1 Architecture Input
2.2 Overall Architecture
2.3 Self-inhibiting Residual Convolutional Block
3 Experiments
3.1 Data and Tasks
3.2 Experiment Settings
3.3 Comparison of Methods
3.4 Result and Discussion
4 Conclusion
References
Recurrent Neural Networks
A Hybrid 2D and 3D Convolution Based Recurrent Network for Video-Based Person Re-identification
1 Introduction
1.1 Motivation
1.2 Contribution
2 Related Works
2.1 Recurrent Neural Networks
2.2 3D Convolutional Networks
3 The Proposed HCRN Network
3.1 Input and Data Augmentation
3.2 3D Convolutional Module
3.3 2D ResBlock Module
3.4 RNN
3.5 Joint Loss Function
4 Experimental Results
4.1 Datasets
4.2 Experimental Settings
4.3 Compared Methods
4.4 Comparison with State-of-the-Art Methods
4.5 Cross Dataset Testing
5 Conclusion
References
Improving Recurrent Neural Networks with Predictive Propagation for Sequence Labelling
1 Introduction
2 Recurrent Neural Networks with Predictive Propagation
2.1 Graphical Structure
2.2 Inference
2.3 Learning
3 Experiments
3.1 OCR
3.2 CoNLL 2000 Chunking
3.3 POS Tagging: Effect of Training Size
3.4 Computational Time
4 Conclusion
References
Design of Synthesizing Multi-valued High-Capacity Auto-associative Memories Based on Complex-Valued Networks
1 Introduction
2 Problem of Descriptions and Preliminaries
3 Stability Analysis and Design Procedures
3.1 Existence and Uniqueness of the Equilibrium Point
3.2 Stability Analysis and Design Procedures
3.3 The Design Procedure of Auto-Associative Memories
4 Numerical Solutions
5 Conclusions
References
Analysis on the Occurrence of Tropical Cyclone in the South Pacific Region Using Recurrent Neural Network with LSTM
Abstract
1 Introduction
2 Model
2.1 Recurrent Neural Networks
2.2 Problem of Long-Term Dependencies
2.3 Long Short-Term Memory
2.4 Structure of the Model Used
3 Factors Causing Formation of Cyclones
3.1 The Current Threshold
3.2 Analysis of the Dataset
4 Analysis and Results
4.1 Testing the Algorithm
4.2 Results
5 Conclusion
References
Combining User-Based and Session-Based Recommendations with Recurrent Neural Networks
Abstract
1 Introduction
2 Related Work
3 Methods
3.1 User and Item Embedding
3.2 RNNs with Modified Gated Recurrent Units
3.3 Training
4 Experiments
4.1 Datasets
4.2 Experimental Setup
4.3 Results
5 Conclusion
Acknowledgement
References
EMD-Based Recurrent Neural Network with Adaptive Regrouping for Port Cargo Throughput Prediction
1 Introduction
2 EMD-Based LSTM with Adaptive Regrouping
2.1 Empirical Mode Decomposition (EMD)
2.2 Standardized Euclidean Distance (SED)
2.3 Long Short-Term Memory (LSTM)
2.4 Three-Step Prediction Framework
3 Experimental Results and Discussion
3.1 Experiment Settings
3.2 Quantitative Performance Evaluation
3.3 Prediction Results and Discussion
4 Conclusions
References
Enhancing the Recurrent Neural Networks with Positional Gates for Sentence Representation
1 Introduction
2 Related Work
3 Position Gated Recurrent Neural Network
3.1 The Framework of PG-RNN
3.2 Positional Interaction Monitor
3.3 Positional Gating Mechanism
4 Experiments
4.1 Experimental Setup
4.2 Effectiveness of Positional Gating Mechanism
4.3 Effectiveness of PG-RNN Model
4.4 Investigation of Propagation Scope
5 Conclusion and Future Work
References
Spiking Neural Networks
A Visual Recognition Model Based on Hierarchical Feature Extraction and Multi-layer SNN
1 Introduction
2 Methods
2.1 Hierarchical Model for Feature Extraction
2.2 Phase Encoding
2.3 Learning Algorithm
3 Simulation Results
3.1 Experimental Setup of Image Coding
3.2 Performance of Multi-layer Learning Rule
3.3 Recognition Performance
4 Conclusion
References
Skewed and Long-Tailed Distributions of Spiking Activity in Coupled Network Modules with Log-Normal Synaptic Weight Distribution
1 Introduction
2 Materials and Methods
2.1 Spiking Neural Network
2.2 Evaluation Indexes
3 Results
4 Discussion and Conclusion
References
Efficient Multi-spike Learning with Tempotron-Like LTP and PSD-Like LTD
1 Introduction
2 Methods
2.1 Neuron Model
2.2 Multi-spike Learning Algorithm
3 Experimental Results
4 Conclusion
References
A Ladder-Type Digital Spiking Neural Network
1 Introduction
2 Digital Spiking Neuron and Periodic Spike-Train
3 Ladder-Type Digital Spiking Neural Network
4 FPGA Based Implementation
5 Conclusions
References
The Effects of Feedback Signals Mediated by NMDA-Type Synapses for Modulating Border-Ownership Selective Neurons in Visual Cortex
Abstract
1 Introduction
2 Methods
2.1 Model Architecture
2.2 Jitter Method for Extracting Tight Synchrony Between BOS Neurons
3 Results
4 Discussion and Conclusion
Acknowledgements
References
Modelling and Analysis of Temporal Gene Expression Data Using Spiking Neural Networks
1 Introduction
2 Materials and Methods
2.1 Data Description
2.2 Gene Expression Data Augmentation and Temporal Feature Selection
2.3 SNN Model
3 Results and Discussion
3.1 Gene Expression Profiling Data Modelling
3.2 Classification
4 Comparative Analysis with Traditional Machine Learning Algorithms
4.1 GIN Analysis
5 Conclusion and Future Work
References
A Gesture Recognition Method Based on Spiking Neural Networks for Cognition Development
1 Introduction
2 Overview of System
3 Gesture Recognition Method Based on SNN
3.1 SNN for Gesture Extraction
3.2 Feature Reprocess
4 Experimental Results
4.1 Gesture Extraction
4.2 Gesture Clustering
4.3 Gesture Classification
4.4 Semantic Association
5 Conclusion
References
Delayed Feedback Reservoir Computing with VCSEL
1 Introduction
2 Experimental Setup
3 Task Description and Results
4 Conclusion
References
Modeling the Respiratory Central Pattern Generator with Resonate-and-Fire Izhikevich-Neurons
Abstract
1 Introduction
2 Modeling Description
2.1 Izhikevich Neurons
2.2 Model Description
3 Results
3.1 rCPG Replicates the Properties of Hodgkin-Huxley-Based Models
3.2 Simulations of Pontine Transection Are Consistent with Experimental Data and Previous Models
3.3 Intrinsically Bursting Properties Are Not Essential to Generate the Three-Phase Respiratory Rhythm
4 Conclusion
Appendix
References
Proposal of Carrier-Wave Reservoir Computing
1 Introduction
2 System Construction
3 Learning Dynamics in Complex Domian
4 Conclusion
References
Spiking Neural Networks for Cancer Gene Expression Time Series Modelling and Analysis
1 Introduction
2 Materials and Methods
2.1 Data Description
2.2 Feature Selection
2.3 SNN Model: The NeuCube Architecture
3 Results and Discussion
3.1 Data Pre-processing and Augmentation
3.2 Feature Selection
3.3 Data Classification with the NeuCube
3.4 Comparison with Support Vector Machine Classifier
3.5 Connectivity Analysis of the NeuCube SNN Model
3.6 GIN Extraction from a Trained NeuCube Model
3.7 Neuronal Connections as Indicators of Gene Ranking by Importance
4 Conclusion and Future Direction
References
Dimensionality Reduction by Reservoir Computing and Its Application to IoT Edge Computing
1 Introduction
2 Relation to Prior Work on Dimensionality Reduction Techniques
3 Random Projection by Reservoir Computing
4 Application to IoT Edge Computing
5 Conclusion
References
Author Index

System requirements

Save as PDF Copy link into clipboard

Schweitzer Fachinformationen

Neural Information Processing

Description

More details

Other editions

Additional editions

Content

System requirements