PRICAI 2018: Trends in Artificial Intelligence

Name: PRICAI 2018: Trends in Artificial Intelligence | 15th Pacific Rim International Conference on Artificial Intelligence, Nanjing, China, August 28-31, 2018, Proceedings, Part I
Brand: Springer
Price: 96.29 EUR
Availability: OnlineOnly

15th Pacific Rim International Conference on Artificial Intelligence, Nanjing, China, August 28-31, 2018, Proceedings, Part I

Xin Geng Byeong-Ho Kang(Editor)

Springer (Publisher)

Published on 30. July 2018

XXIX, 1086 pages

E-Book

PDF with digital watermarking

System requirements

978-3-319-97304-3 (ISBN)

€96.29incl. 7% vat

System requirements

for PDF with digital watermarking

E-Book Single Licence

Available for download

Description

More details

Other editions

Content

Intro
Preface
Organization
Contents - Part I
Contents - Part II
HAVAE: Learning Prosodic-Enhanced Representations of Rap Lyrics
1 Introduction
2 Related Work
3 Model Description
3.1 Preliminary
3.2 Feature Extraction Module
3.3 Feature Aggregation Module
4 Experiments and Results
4.1 NextLine Prediction
4.2 Rap Genre Classification
5 Conclusions
References
DKE-RLS: A Manifold Reconstruction Algorithm in Label Spaces with Double Kernel Embedding-Regularized Least Square
1 Introduction
2 Related Work
3 A Double Kernel Embedding-Regularized Least Square Method (DKE-RLS) Algorithm
3.1 Transformation of Local Topologies from Feature Space to Label Space
3.2 Using a Kernel Matrix in the Feature Space
4 Experiments
4.1 Experimental Setup
4.2 Experiment Results
4.3 Time Complexity
5 Conclusion
References
Learning Relations from Social Tagging Data
1 Introduction
2 Related Work
3 A Supervised Model for Learning Tag Relations
3.1 Probabilistic Topic Analysis of Tagging Data
3.2 Assumptions for Feature Set Generation
3.3 Feature Set Generation
4 Experimental Results and Evaluation
5 Conclusion and Future Work
References
Selecting Optimal Source for Transfer Learning in Bayesian Optimisation
1 Introduction
2 Preliminaries
2.1 Gaussian Process-1mm
2.2 Bayesian Optimisation-1mm
2.3 Multi-armed Bandit Problem-1mm
2.4 Transfer Learning for Bayesian Optimisation
3 Proposed Method
3.1 Source/Target Relatedness
3.2 Source Selection Strategy
4 Experiments
4.1 Experimental Setting-2mm
4.2 Synthetic Experiments-2mm
4.3 Hyperparameter Tuning -1mm
4.4 Short Polymer Fiber Synthesis-1mm
5 Conclusion
References
Fast Spatially-Regularized Correlation Filters for Visual Object Tracking
1 Introduction
2 Background
3 The Method
3.1 Fast Spatial Regularization for CF
3.2 Fast Online Learning of SRCFNet
3.3 SRCFNet Boosted CF Tracking
4 Experimental Results
4.1 Setup
4.2 SRCFNet Based CF Trackers
4.3 Comparison with State-of-the-Art Trackers
4.4 Discussion About SR Weight Map W
5 Conclusion
References
Similarity-Adaptive Latent Low-Rank Representation for Robust Data Representation
Abstract
1 Introduction
2 Related Work: LatLRR
3 Similarity-Adaptive Latent Low-Rank Representation (SA-LatLRR) for Robust Data Representation
3.1 Proposed Formulation
3.2 Optimization
3.3 Discussion: Relationship Analysis
4 Experimental Results and Analysis
4.1 Visualization of Learnt Representation Coefficients Z
4.2 Image Representation by Decomposition
4.3 Quantitative Evaluation of Image De-Noising
4.4 Face Recognition by Feature Extraction
5 Conclusion Remarks
Acknowledgment
References
Adaptively Shaping Reinforcement Learning Agents via Human Reward
1 Introduction
2 Reinforcement Learning
3 Human-Agent RL Methods
4 Adaptively Shaping RL Agents
5 Experimental Evaluations
5.1 Experimental Setup
5.2 Experimental Results
6 Related Work
7 Conclusions
References
Incomplete Multi-view Clustering via Structured Graph Learning
1 Introduction
2 The Proposed SGL
3 Optimization Algorithms
4 Discussion
4.1 Convergence Analysis
4.2 Computational Time
5 Experiment
5.1 Data Sets
5.2 Comparing Methods
5.3 Clustering Result Comparison
5.4 Convergence Study
5.5 Parameter Study
6 Conclusion
References
DeepRSD: A Deep Regression Method for Sequential Data
1 Introduction
2 Designs of DeepRSD
2.1 RNN Review
2.2 The Overall Architecture
2.3 Activation Function
2.4 Alternative Dropout
3 Training and Inference
4 Experiment
4.1 Evaluations on AMS Solar Energy Prediction Contest
4.2 Evaluations on Electricity Demand Forecasting Competition
5 Conclusion
References
Single Image Super-Resolution via Perceptual Loss Guided by Denoising Auto-Encoder
1 Introduction
2 Related Work
3 The Proposed Method
3.1 Motivation
3.2 The Architecture
3.3 Denoising Auto-Encoder
3.4 Perceptual Loss Function
4 Experiments
4.1 Training Details and Parameters
4.2 Results
5 Conclusion
References
Context-Aware Phrase Representation for Statistical Machine Translation
1 Introduction
2 Related Work
3 The TBRAE Model
3.1 RAE-Based Phrase Modeling
3.2 Context Modeling
3.3 Bilingually-Constrained Topical Phrase Embeddings
3.4 Word-Topic Semantic Constraint Modeling
3.5 The Model Objective
4 Experiments
4.1 Setup
4.2 Overall Performance
5 Conclusions and Future Work
References
Collaborating Aesthetic Change and Heterogeneous Information into Recommender Systems
1 Introduction
2 Related Work
3 Framework
3.1 Problem Definition
3.2 Overview
3.3 Preprocessing Part
3.4 User Knowledge Representation Part
3.5 Item Knowledge Representation Part
3.6 Output Layers
3.7 Training
4 Experiment
4.1 Datasets and Metrics
4.2 Baselines
4.3 Results
5 Conclusions
References
Latent Subspace Representation for Multiclass Classification
1 Introduction
2 The Proposed Method
2.1 Motivation
2.2 Problem Formulation
3 Optimization
3.1 Optimization for LSRMC
3.2 Complexity and Convergence
4 Experiments
4.1 Experiment Setting
4.2 Experimental Results
4.3 Parameter Tuning and Convergence Experiment
5 Conclusions
References
Low-Rank Graph Regularized Sparse Coding
Abstract
1 Introduction
2 LogSC: Low-Rank Graph Regularized Sparse Coding
2.1 Traditional Sparse Coding (TSC)
2.2 Laplacian Regularization
2.3 Low-Rank Representation (LRR)
2.4 Objective Problem of LogSC
3 Objective Optimization by LADMAP
3.1 Updating E
3.2 Updating Z
3.3 Updating W
3.4 Updating X
3.5 Updating D
3.6 Updating M1 and M2
4 Experiments
4.1 Image Clustering on Yale
4.2 Image Classification on AR
4.3 Robust Image Classification on USPS and COIL20
5 Conclusion
Acknowledgments
References
Decentralized Multiagent Reinforcement Learning for Efficient Robotic Control by Coordination Graphs
1 Introduction
2 RL, MARL and CG
2.1 MDP and RL
2.2 MARL and CG
3 Decentralized MARL for Robotic Control
3.1 Robot Decomposition Using CG
3.2 Problem Formalization
3.3 Computation of the Continuous Joint Optimal Actions
4 Experiments
5 Related Work
6 Conclusions
References
Construction of Microblog-Specific Chinese Sentiment Lexicon Based on Representation Learning
Abstract
1 Introduction
2 Related Work
3 The Proposed Approach
3.1 Microblog-Specific Word Embedding
3.2 Lexicon Construction
4 Experimental Evaluation
4.1 Corpus and Settings
4.2 Evaluation of Lexicon
4.3 Evaluation of Feature Learning
5 Conclusion
Acknowledgment
References
Phonologically Aware BiLSTM Model for Mongolian Phrase Break Prediction with Attention Mechanism
1 Introduction
2 Proposed Model
2.1 Input Features
2.2 Attention Mechanism
2.3 BiLSTM Phrase Break Model
3 Experiments and Analysis
3.1 Datasets
3.2 Setup
3.3 Main Results
3.4 Comparison of Phonological Embeddings Dimensions
3.5 Comparison of Position-Based Enhanced Method
3.6 Comparison of Output Layer
3.7 Listening Tests
4 Conclusions
References
Multi-label Crowdsourcing Learning with Incomplete Annotations
1 Introduction
2 Related Work
3 Crowdsourcing with Incomplete Annotations
3.1 Missing Annotation Estimation
3.2 Groundtruth Inference
4 Active Annotation Collection
5 Running Time Analysis
6 Experiment
6.1 Data Sets
6.2 Crowdsourcing Learning
6.3 Active Results
6.4 Parameter Study
7 Conclusion
References
Multiple Kernel Fusion with HSIC Lasso
Abstract
1 Introduction
2 Multiple Kernel Learning
3 MKL-HSIC
4 Experimental Evaluation
5 Conclusion
Acknowledgements
References
Visualizing and Understanding Policy Networks of Computer Go
Abstract
1 Introduction
2 Related Work
3 Network
3.1 Data
3.2 Network Architecture
3.3 Training Results
4 Visualization
4.1 Visualize the Weight of Filters
4.2 Visualize the Activation
4.3 Visualization with a Transposed Convolution Network
4.4 Occlusion Experiment
5 Conclusion and Future Work
References
A Multi-objective Optimization Model for Determining the Optimal Standard Feasible Neighborhood of Intelligent Vehicles
Abstract
1 Introduction
2 Related Works
3 The Multi-objective Optimization Model for the Standard Feasible Neighborhood
3.1 The Definition of the Multi-objective Optimization Model
3.2 The Proposed Algorithm for the Multi-objective Optimization Model
3.3 An Illustrative Example of the Partition Method
4 Performance Evaluation
4.1 Comparison on Different Road Conditions
4.2 Comparison of Different Neighborhood Models
5 Conclusion
References
Efficient Detection of Critical Links to Maintain Performance of Network with Uncertain Connectivity
1 Introduction
2 Related Work
3 Problem Formulation
4 Proposed Method
4.1 RCS: Reachability Condition Skipping
4.2 DCS: Distance Constraints Skipping
5 Experiments
5.1 Evaluation of Computational Efficiency
5.2 Evaluation of Contribution Values
5.3 Visualization of Detected Link Locations
6 Conclusion
References
Mixed Neighbourhood Local Search for Customer Order Scheduling Problem
1 Introduction
2 Preliminaries
3 Related Works
4 Our Approach
4.1 The Proposed RBM Heuristic
4.2 Search Algorithm
5 Experimental Results
6 Conclusion
References
Graph Based Family Relationship Recognition from a Single Image
1 Introduction
2 Related Work
3 Proposed Method
3.1 Face Feature Extraction
3.2 Pairwise Kinship Classification
3.3 Family Relationship Network Construction
4 Experiment
4.1 Databases
4.2 Result
5 Conclusions
References
ACGAIL: Imitation Learning About Multiple Intentions with Auxiliary Classifier GANs
1 Introduction
2 Background
2.1 MDPs
2.2 Single-Intention Imitation Learning
2.3 Multiple-Intention Imitation Learning
3 ACGAIL: Imitation Learning with Auxiliary Classifier
3.1 ACGAIL
3.2 Algorithm Implementation
3.3 Relation to InfoGAIL and Mutual Information
4 Experiments
4.1 Environmental Setup
4.2 Imitation Learning About Multiple Intentions
5 Conclusion and Future Work
References
Matching Attention Network for Domain Adaptation Optimized by Joint GANs and KL-MMD
1 Introduction
2 Related Work
2.1 Domain Transfer Learning
2.2 Attention Mechanism
3 The Proposed Method
3.1 Model Architecture
3.2 Training and Testing
4 Expriments
4.1 Setup and Results
4.2 Analysis
5 Conclusion
References
Attention Based Meta Path Fusion for Heterogeneous Information Network Embedding
1 Introduction
2 Related Work
3 Preliminaries
4 The Proposed Method
4.1 Meta Path Based Homogeneous Network
4.2 Homogeneous Network Embedding via AutoEncoder
4.3 Fusing Embeddings via Attention Mechanism
5 Experiments
5.1 Datesets
5.2 Baselines
5.3 Parameter Settings
5.4 Classification
5.5 Clustering
5.6 Analysis of Attention Mechanism
5.7 Visualization
5.8 Parameters Experiments
6 Conclusion
References
An Efficient Auction with Variable Reserve Prices for Ridesourcing
1 Introduction
1.1 Our Results
2 Related Work
3 Preliminary
3.1 Discussion on VCG-type Mechanisms
3.2 Interpretation in Ridesourcing
4 A Truthful Approximation Mechanism
4.1 Truthfulness
4.2 Individual Rationality and Reserve Price Constraints
4.3 Approximation Ratio
4.4 Lower Bounds
5 Evaluation
5.1 Simulation Setup
5.2 Benchmark Mechanisms
5.3 Results
5.4 Discussion
6 Conclusion and Future Work
References
Matrix Entropy Driven Maximum Margin Feature Learning
1 Introduction
2 Related Work
3 Robust Maximum Margin Framework
3.1 Rationality of the 1-norm
3.2 Maximum Margin Framework
4 Matrix Entropy Driven Maximum Margin Feature Learning
4.1 Theoretical Analysis
4.2 M3FL Approach
5 Discussion
6 Experiments
6.1 Results on Gene Expression Data
6.2 Results on Image Data
7 Conclusions and Future Work
References
Spectral Image Visualization Using Generative Adversarial Networks
1 Introduction
2 Generative Adversarial Nets
3 Proposed Method
3.1 Formulation
3.2 Architecture
4 Experiments
4.1 Datasets
4.2 Implementation Details
4.3 Visual and Quantitative Comparisons
5 Conclusion
References
Fusing Semantic Prior Based Deep Hashing Method for Fuzzy Image Retrieval
1 Introduction
2 Related Works
3 Approach
3.1 Problem Statement
3.2 Deep Architecture
3.3 FSPDH Framework
3.4 Optimization
4 Experiment
4.1 Dataset
4.2 Experimental Results
4.3 Parameter Sensitivity
5 Conclusion
References
Topic-Guided Automatical Human-Simulated Tweeting System
1 Introduction
2 Related Work
2.1 Analysis Systems on Tweet Data
2.2 Image Caption
3 Human-Simulated Tweeting System
3.1 Keyword-Based Retrieval Module
3.2 Topic-Guided Image Captioning Module
3.3 System Mechanism
4 Topic-Guided Image Captioning Module
4.1 Memory Component
4.2 Text Generation
5 Experiments
5.1 Dataset and Experimental Setting
5.2 Quantitative Comparison
5.3 Qualitative Analysis
5.4 System Performance
6 Conclusions
References
Network Embedding Based on a Quasi-Local Similarity Measure
1 Introduction
2 DeepWalk and Node2Vec
3 New Objective Function
4 Similarity Measure
5 The Algorithm
6 Experiments
6.1 Multilabel Classification
6.2 Link Prediction
7 Related Work
8 Conclusion
References
Reinforcement Learning for Mobile Robot Obstacle Avoidance Under Dynamic Environments
1 Introduction
2 The Environment Model
2.1 The State Space
2.2 The Action Space
3 Collision Avoidance Using Improved Q-Learning
3.1 The Reward Function
3.2 The Q-Values
3.3 The Selection Policy
4 Simulation and Analysis
4.1 Some Test Simulations
4.2 Comparison Results
5 Conclusions and Future Work
References
Subclass Maximum Margin Tree Error Correcting Output Codes
1 Introduction
2 Related Encoding Algorithms
3 Subclass Maximum Margin Tree ECOC (SM2ECOC)
4 Experiments and Analyses
5 Conclusion
References
A Multi-latent Semantics Representation Model for Mining Tourist Trajectory
Abstract
1 Introduction
2 Related Works
2.1 Spatio-Temporal Trajectory Mining
2.2 Travel Attraction Recommendation
3 A Multi-latent Semantics Tourist Trajectory Representation Model
3.1 Framework of the Model
3.1.1 User-Level Latent Semantics
3.1.2 Temporal-Level Latent Semantics
3.1.3 Attractions-Level Latent Semantics
3.2 Modeling
3.3 Parameter Estimation and Optimization
3.4 Model Application
3.5 Datasets
4 Experiment
4.1 Parameter Training
4.2 Latent Semantic Analysis of Model
4.3 Evaluation Criterion
4.4 Comparison Experiment
5 Conclusion
Acknowledgments
References
Two-Stage Unsupervised Deep Hashing for Image Retrieval
1 Introduction
2 Related Work
2.1 Hashing Techniques
2.2 Deep Learning
3 The Proposed Approach
3.1 The First-Stage Training
3.2 The Second-Stage Training
4 Experiments
4.1 Datasets
4.2 Evaluation Metric
4.3 Compared Methods
4.4 Implementation Details
4.5 Results
5 Conclusion
References
A Fast Heuristic Path Computation Algorithm for the Batch Bandwidth Constrained Routing Problem in SDN
1 Introduction
2 Problems Definition and Related Works
2.1 The Bandwidth Constrained Routing
2.2 The Graph Partitioning
3 A Fast Path Computation Algorithm
3.1 Routing Algorithm on Quotient Graph
3.2 The Path Computation Time
4 Performance Evaluation
4.1 Graphs and Request Sets
4.2 Results
5 Conclusion
6 Further Discussion
References
3SP-Net: Semantic Segmentation Network with Stereo Image Pairs for Urban Scene Parsing
1 Introduction
2 Related Work
3 Methodology
3.1 Combine RGB Features and Depth Features
3.2 Discriminative Network
3.3 Adversarial Training
3.4 Network Architectures
4 Experiments
4.1 Results of 3SP-Net Based on FCN
4.2 Results of 3SP-Net Based on Other Networks
5 Conclusions
References
An Interactivity-Based Personalized Mutual Reinforcement Model for Microblog Topic Summarization
Abstract
1 Introduction
2 Related Work
2.1 Traditional Text Summarization
2.2 Microblog Summarization
3 Our Model
3.1 Extracting Keywords and Candidate Sentences
3.2 Ranking Candidate Sentences Based on Personalized Mutual Reinforcement Model
4 Experimental Settings
4.1 Datasets
4.2 Evaluation Metric
4.3 Baselines
5 Experimental Results
5.1 Testing the Relevance Between Interactivity and 'Elite Posts'
5.2 Comparison with Baselines
5.3 Parameter Sensitivity Analysis
6 Conclusion and Future Work
Acknowledgments
References
Multiple Visual Fields Cascaded Convolutional Neural Network for Breast Cancer Detection
1 Introduction
2 Small Visual Field CNN
2.1 Slide Preprocessing
2.2 Patch Sampling
2.3 Relatively Hard Example Mining
3 Large Visual Field CNN
4 Multiple Visual Fields Cascaded CNN
5 Experimental Evaluation
5.1 Evaluation of Different Tumor Patch Definitions
5.2 Evaluation of Hard Example Mining
5.3 Evaluation of Multiple Visual Fields Cascaded CNN
6 Conclusion
References
Multi-view Learning and Deep Learning for Microscopic Neuroblastoma Pathology Image Diagnosis
1 Introduction
2 Related Work
3 Traditional Representations and Deep Representations
4 Multi-view Learning with Maximum Entropy Discrimination
5 Experiments
5.1 Dataset Description
5.2 Binary Classification
5.3 Multi-class Classification
6 Conclusion
References
Low-Rank Matrix Recovery via Continuation-Based Approximate Low-Rank Minimization
1 Introduction
2 Related Works
3 The Proposed Model
3.1 Problem Formulation
3.2 Optimization
4 Convergence Analysis
5 Experiments
5.1 Synthetic Data
5.2 Real-World Data
6 Conclusion
References
Inertial Constrained Hierarchical Belief Propagation for Optical Flow
1 Introduction
2 Method
2.1 General Framework of BP
2.2 Inertial Constrained Hierarchical BP Model
2.3 Postprocessing
3 Experiments
3.1 Parameter Analysis
3.2 Inertial Information Analysis
3.3 Experiments on MPI-Sintel
4 Conclusion
References
ParallelNet: A Depth-Guided Parallel Convolutional Network for Scene Segmentation
1 Introduction
2 Related Work
3 Methodology
3.1 The Depth Determination Module (DDM)
3.2 Cascaded Depth and RGB Features Fusion Block (CFB)
3.3 Other Network Variants
4 Experiments
4.1 Comprehensive Experiments
4.2 Ablation Studies
4.3 Qualitative Results
5 Conclusion
References
Aircraft Detection in Remote Sensing Images Based on Background Filtering and Scale Prediction
1 Introduction
2 Related Work
3 Our Method
3.1 Background-Filtering-Network to Crop Images
3.2 Scale-Faster-R-CNN
4 Experiments
5 Conclusion
References
Residual Convolutional Neural Networks with Global and Local Pathways for Classification of Focal Liver Lesions
Abstract
1 Introduction
2 Proposed Method
2.1 Global Pathway
2.2 Local Pathway
2.3 Joint Loss Function
2.4 Post-processing of Label Map and Classification of Lesions by SVM
2.5 Extension to Multi-phase CT Images
3 Experiments and Results
3.1 Setup
3.2 Results
4 Conclusion
Acknowledgements
References
Intent Detection for Spoken Language Understanding Using a Deep Ensemble Model
1 Introduction
2 Related Works
3 Methodology
3.1 Word Embedding Representation
3.2 Baseline Models
3.3 Proposed Approach
4 Dataset and Experiments
4.1 Dataset
4.2 Experimental Setup
5 Results and Discussions
5.1 Results
5.2 Analysis
5.3 Statistical Test
6 Conclusion and Future Work
References
Accurately Detecting Community with Large Attribute in Partial Networks
Abstract
1 Introduction
2 Related Works
3 Model
3.1 Modeling Network Relations and Attributes
3.2 The Model with Regularization Constraints
4 Optimization
5 Experiments
5.1 Datasets
5.2 Baseline and Evaluation Metrics
5.3 Results
6 Conclusions
Acknowledgements
References
Two-Step Multi-factor Attention Neural Network for Answer Selection
1 Introduction
2 Two-Step Multi-factor Attention Model
2.1 Question and Answer Encoder Layer
2.2 Attention Layer
2.3 Loss Function
3 Experiments
3.1 Datasets
3.2 Model Comparison
3.3 Experimental Settings
3.4 Experimental Results and Analysis
4 Model Properties
5 Related Work
6 Conclusion
References
Labeling Information Enhancement for Multi-label Learning with Low-Rank Subspace
1 Introduction
2 Related Work
3 The LIEML Algorithm
3.1 Label Enhancement
3.2 Predictive Model Induction
4 Experiments
4.1 Experiment Configuration
4.2 Experimental Results
5 Conclusion
References
Deep Coordinated Textual and Visual Network for Sentiment-Oriented Cross-Modal Retrieval
1 Introduction
2 Related Work
2.1 Traditional Methods
2.2 Deep Learning Based Methods
3 Methodology
3.1 Visual Branch
3.2 Textual Branch
3.3 Coordinated Feature Learning
4 Experiment
4.1 Dataset
4.2 Experiment Settings
4.3 Baseline
4.4 Evaluation Metrics
4.5 Results Analysis
4.6 Visualization
5 Conclusion
References
A New Context-Based Clustering Framework for Categorical Data
1 Introduction
2 Related Work
3 The Proposed Clustering Algorithm
3.1 Notations
3.2 k-Means Like Clustering Framework
3.3 A Context-Based Dissimilarity Measure for Categorical Data
4 Experimental Evaluation
4.1 Testing Datasets
4.2 Clustering Quality Evaluation
4.3 Experimental Results
5 Conclusion
References
TypicFace: Dynamic Margin Cosine Loss for Deep Face Recognition
Abstract
1 Introduction
2 Related Work
2.1 Normalization Approaches
2.2 Softmax-Based Loss Functions
3 From Softmax to TypicFace
3.1 Softmax
3.2 A-Softmax [2]
3.3 AM-Softmax [3]
3.4 TypicFace (Proposed)
4 Periments
4.1 Implementation Details
4.2 Experiments on LFW [4] and BLUFR [5]
5 Conclusions
6 Future Work
References
Semi-supervised Feature Selection Based on Logistic I-RELIEF for Multi-classification
1 Introduction
2 Related Work
2.1 Logistic I-RELIEF
2.2 Semi-supervised Logistic I-RELIEF
3 Semi-supervised Logistic I-RELIEF for Multi-classification
3.1 Calculation of Margin Vectors
3.2 Optimization Problem and Algorithm Description
4 Experiments
4.1 Artificial Dataset
4.2 UCI Datasets
5 Conclusion
References
Genetic Programming for Feature Selection and Feature Construction in Skin Cancer Image Classification
1 Introduction
2 Background
2.1 Related Work
2.2 Local Binary Patterns
3 The Proposed Method
3.1 Fitness Function
3.2 Terminal Set and Function Set
4 Experiment Design
4.1 Dataset
4.2 GP Parameters
4.3 Methods for Classification
5 Results and Discussions
5.1 Overall Results
5.2 Analysis of the Evolved Features
6 Conclusions
References
Unsupervised Stereo Matching with Occlusion-Aware Loss
1 Introduction
2 Related Work
3 Proposed Scheme
3.1 Framework
3.2 Neural Network Architecture
3.3 Occlusion Map
3.4 Loss Functions
4 Experiment
4.1 Setup
4.2 Comparisons
4.3 Effect of Occlusion-Aware Reconstruction Loss
5 Conclusion
References
Siamese Network Based Features Fusion for Adaptive Visual Tracking
1 Introduction
2 Related Work
3 The Proposed Approach
3.1 Siamese Network Framework
3.2 Correlation Filter
3.3 Feature Fusion
3.4 Visual Object Tracking Algorithm
4 Experiments
4.1 Evaluation of Different Convolutional Layer Depths
4.2 Comparisons with State-of-the-Art Methods
5 Conclusion
References
ANNC: AUC-Based Feature Selection by Maximizing Nearest Neighbor Complementarity
1 Introduction
2 Feature Selection Based on AUC by Maximizing Nearest Neighbor Complementarity (ANNC)
2.1 Feature Complementarity in ANNC
2.2 Procedure of ANNC
2.3 Evaluation of K
3 Experiments
3.1 Exp. 1: Noisy Feature Test
3.2 Exp. 2: Classification Performance
3.3 Exp. 3: Evaluation of
4 Conclusion
References
Prediction of Nash Bargaining Solution in Negotiation Dialogue
1 Introduction
2 Related Work
2.1 Automated Negotiation (Agent-Agent Negotiation)
2.2 Negotiation Agent for Negotiation Dialogues in Natural Language
2.3 Recurrent Neural Networks (RNNs) for Natural Language Processing
3 Problem Definition
4 Prediction of Nash Bargaining Solution
5 Evaluation
5.1 Experimental Settings
5.2 Experimental Results
6 Conclusion
References
Joint Residual Pyramid for Depth Map Super-Resolution
1 Introduction
2 Background and Related Work
2.1 Local-Based Methods
2.2 Global-Based Methods
2.3 CNN-Based Methods
3 Proposed Method
3.1 Overview
3.2 Convolutional Neural Residual Pyramid Network
3.3 Joint SR Network
3.4 Loss Function
4 Experiments and Applications
4.1 Training Settings
4.2 Pyramid Levels vs. Performance
4.3 Loss Functions and Network Structures
4.4 Depth Map SR
4.5 Chromaticity Map SR
4.6 Saliency Map SR
5 Conclusion and Future Work
References
Reading More Efficiently: Multi-sentence Summarization with a Dual Attention and Copy-Generator Network
1 Introduction
2 Related Work
3 Models
3.1 Sequence to Sequence Text Summarization
3.2 Dual Attention
3.3 Copy-Generator Network
4 Experiments
4.1 Dataset
4.2 Evaluation Metric
4.3 Implementation Details
4.4 Results
5 Conclusion
References
Staged Generative Adversarial Networks with Adversarial-Boundary
1 Introduction
2 Related Works
3 Preliminaries
3.1 Generative Adversarial Networks
3.2 AC-GAN
3.3 WGANs
4 Methods
4.1 StageGAN
4.2 AB-GAN
4.3 ABS-GAN
5 Experiments
5.1 CIFAR-10
5.2 STL-10
5.3 CelebA
6 Conclusion and Future Work
References
Semi-supervised DenPeak Clustering with Pairwise Constraints
1 Introduction
2 Related Work
3 Our Method
3.1 Generating Temporary Clusters
3.2 Merging Temporary Clusters
4 Experiments
4.1 Experimental Setup
4.2 Results on Synthetic Data Sets
4.3 Results on Real Data Sets
4.4 Sensitivity Analysis
4.5 Time Complexity Analysis
5 Conclusion and Future Work
References
A Novel Convolutional Neural Network for Statutes Recommendation
Abstract
1 Introduction
2 Background
2.1 Recommender System
2.2 Convolutional Neural Network
2.3 Attention Mechanism
2.4 Word Vectors
3 The Proposed CNN Structure
3.1 Correlation Matrix
3.2 Convolution
3.3 K-max Pooling
3.4 Full Connection
4 Experiments and Results
4.1 Datasets
4.2 Pre-trained Word Vectors
4.3 Training
4.4 Results
5 Conclusion
Acknowledgment
References
Towards Understanding User Requests in AI Bots
1 Introduction
2 Related Work
3 A Proposed Approach to Analyze User Requests
3.1 Analyzing User Requests as a Two-Layer Sequence Labeling Task
3.2 A Proposed Models for Solving the Sequence Labeling Task
4 Experiments
4.1 Building the Corpus
4.2 Experimental Setups
4.3 Experimental Results
5 Error Analysis
6 Conclusion
References
A Deep Reinforced Training Method for Location-Based Image Captioning
1 Introduction
2 Related Work
3 Methodology
3.1 Encoder-Decoder Framework
3.2 Location-Based Mechanism for Image Encoding
3.3 Attention-Based LSTM for Decoding
3.4 Combined Training Method
4 Experiments
4.1 Datasets and Settings
4.2 Implementation Details
4.3 Performance on Datasets
4.4 Qualitative Analysis
5 Conclusion
References
Graph Stream Mining Based Anomalous Event Analysis
1 Introduction
2 Problem Statement
3 Methodology
3.1 Pedestrian Monitoring
3.2 Graph Representation
3.3 Anomaly Detection
4 Results and Discussion
4.1 Datasets and Experimental Setup
4.2 Quantitative Analysis and Evaluation on Real-World Datasets
4.3 Quantitative Analysis and Evaluation on UCSD Datasets
4.4 Discussion
5 Conclusions
References
Nonlinearized Relevance Propagation
1 Introduction
2 Previous Work
2.1 Sensitivity Analysis
2.2 Layer-Wise Relevance Propagation and -rule
2.3 Special Relevance Propagation Rules
3 Proposed Methods
3.1 Nonlinear Functions
3.2 Propagation Rules
3.3 Application to Attentive Pooling Network
4 Experiments
4.1 Dataset and Preprocessing
4.2 Model Setup
4.3 Validation of Relevance
5 Conclusion
References
Online Personalized Next-Item Recommendation via Long Short Term Preference Learning
1 Introduction
2 Related Work
3 Long-Term and Short-Term Matrix Factorization
3.1 Problem Definition
3.2 Long- and Short-Term Preference Model (LSPM)
3.3 Parameters Learning
3.4 Updating Strategy
4 Experiment
4.1 Experimental Setup
4.2 Leave-One-Out Protocol
4.3 Updating Protocol
4.4 Parameter Analysis
5 Conclusion
References
Enhancing Artificial Bee Colony Algorithm with Superior Information Learning
Abstract
1 Introduction
2 Related Work
2.1 The Basic ABC Algorithm
2.2 The MABC-NS Algorithm
3 Our Approach
3.1 Superior Information Learning (SIL) Strategy
3.2 SILABC Algorithm
4 Experimental Results
4.1 Benchmark Functions and Parameter Settings
4.2 Effectiveness of the Proposed Approach
4.3 Comparison with Several up-to-Date ABC Variants
5 Conclusion
Acknowledgments
References
Robust Factorization Machines for Credit Default Prediction
1 Introduction
2 Preliminary
2.1 Credit Default Prediction
2.2 Factorization Machines
3 Smooth Asymmetric Ramp Loss
4 Parameter Estimation
5 Experiments
5.1 Experimental Settings
5.2 Performance Evaluation
5.3 Hyper-parameter Study
6 Related Work
7 Conclusion and Future Work
References
Multi-label Active Learning with Conditional Bernoulli Mixtures
1 Introduction
2 Bernoulli Mixtures and Conditional Bernoulli Mixtures
3 Methods
3.1 Learning Framework
3.2 Selection Criteria
3.3 Sampling Bias Correction
4 Experiments
4.1 Experimental Setting
4.2 Datasets and Results
4.3 Discussion
5 Conclusion and Future Work
References
Investigating the Dynamic Decision Mechanisms of Users' Relevance Judgment for Information Retrieval via Log Analysis
1 Introduction
2 Related Work
2.1 The Enriched Multidimensional User Relevance Model
2.2 Interaction Between Relevance Dimensions
3 Research Questions and Experimental Design
3.1 Research Questions and Corresponding Hypotheses
3.2 Experimental Design
4 Experimental Results and Analysis
4.1 Order Effects Between Dimensions
4.2 Sequential Effects in Document Relevance Judgment
4.3 A Further Study on TREC Session Track Task
5 Conclusion
References
A Correlation-Aware ML-kNN Algorithm for Customer Value Modeling in Online Shopping
1 Introduction
2 Research Background
2.1 Customer Value Modeling
2.2 RFM Model
2.3 Multi-label Learning
3 CAML-kNN Model
3.1 The Basic ML-kNN Model
3.2 Model Development
4 Experimental Results
4.1 The Experimental Data
4.2 Experiments on Customer Labeling
4.3 Evaluation Metrics on Multi-label Learning
4.4 Experiments on Multi-label Learning
5 Conclusion
References
Binary Collaborative Filtering Ensemble
1 Introduction
2 The Proposed BCFE Approach
2.1 Preliminaries
2.2 Problem Formulation
2.3 Solution
3 Experimental Analysis
3.1 Experimental Setup
3.2 Result Analysis
4 Related Work
5 Conclusions
References
Social Collaborative Filtering Ensemble
1 Introduction
2 Related Work
3 The Proposed SoTriCF Approach
3.1 Preliminaries
3.2 Ensembling Three CF Methods
3.3 Exploiting Social Signal in TriCF
3.4 Enhancing the Latent Representation of Cold-Start Users
4 Experimental Analysis
4.1 Experimental Setup
4.2 Result Analysis
5 Conclusions
References
High-Performance OCR on Packing Boxes in Industry Based on Deep Learning
Abstract
1 Introduction
2 Relative Work
3 Ours Method
3.1 Character Regions Detection
3.2 Column Classification
3.3 Single Character Recognition
4 Experiments
4.1 Dataset and Evaluation Protocol
4.2 Experimental Results
5 Discussion
6 Conclusion
Acknowledgement
References
An Implementation of Large-Scale Holonic Multi-agent Society Simulator and Agent Behavior Model
1 Introduction
2 Conceptual Architecture: A Large-Scale Holonic Multi-agent Simulator
2.1 Outline Hadfi2017its,Hadfi2016aamas
2.2 The Architecture
3 Behavior Modeling
3.1 Driver Behavior Modeling
3.2 Extracting Model Parameters from Driving Data
4 Distributed Implementation
4.1 Distributed Simulator
4.2 3D Visualization
5 Conclusions
References
Establishing Connections in a Social Network
1 Introduction
2 A Typology of Centrality
3 Algorithms for the Network Building Problem
4 Experiment and Analysis
5 Conclusion and Future Work
References
Gradient Hyperalignment for Multi-subject fMRI Data Alignment
Abstract
1 Introduction
2 Hyperalignment
3 Gradient Hyperalignment
3.1 Optimization
4 Experiments
4.1 Task Analysis
4.2 Classification by Using Feature Selection
4.3 Classification by Changing the Batch Size
4.4 Classification by Changing the Iterations
4.5 Runtime Analysis
5 Conclusion
Acknowledgments
References
Node Based Row-Filter Convolutional Neural Network for Brain Network Classification
Abstract
1 Introduction
2 Method
2.1 Architecture of NRF-CNN
2.2 Definition of Graph Local Pattern
2.3 Node Based Row-Filter Convolution
2.4 Structure Preserved Pooling
2.5 Feature Fusion with Different Layers
3 Experiments
3.1 Data Collection
3.2 Compared Methods
3.3 Experiment Setting
3.4 Performance on Brain Disease Classification Problem
3.5 Performance with Data Augmentation
3.6 Lesion Detection of Brain Region
3.7 Feature Representations from Different Layers
4 Conclusion
Acknowledgements
References
Author Index

System requirements

Save as PDF Copy link into clipboard

Schweitzer Fachinformationen

PRICAI 2018: Trends in Artificial Intelligence

Description

More details

Other editions

Additional editions

Content

System requirements