Intro
Preface
Organization
Contents
Part I
Human-Computer Interaction
A Genetic Feature Selection Based Two-Stream Neural Network for Anger Veracity Recognition
1 Introduction
2 Method
2.1 Dataset
2.2 Network Architecture
2.3 Two-Stream Architecture
2.4 Data Pre-processing and Feature Selection
3 Experiments and Discussions
3.1 Experiment Settings
3.2 Baseline Model
3.3 Experiments on GFS and Two-Stream Architecture
3.4 Discussion
4 Conclusion and Future Work
References

An Efficient Joint Training Framework for Robust Small-Footprint Keyword Spotting
1 Introduction
2 System Description
2.1 Masking-Based Speech Enhancement Method
2.2 Feature Transformation Block
2.3 Keyword Spotting System
3 Experiments and Results
3.1 Experimental Settings
3.2 Results
4 Conclusions
References
Hierarchical Interactive Matching Network for Multi-turn Response Selection in Retrieval-Based Chatbots
1 Introduction
2 Related Work
3 Hierarchical Interactive Matching Network
3.1 Task Description
3.2 Model Overview

3.3 Multi-level Attention Representation
3.4 Two-Level Hierarchical Interactive Matching
3.5 Aggregation
4 Experiments
4.1 Dataset
4.2 Evaluation Metric
4.3 Baseline Models
4.4 Experiment Settings
4.5 Experiment Results
4.6 Discussions
5 Conclusion
References
Investigation of Effectively Synthesizing Code-Switched Speech Using Highly Imbalanced Mix-Lingual Data
1 Introduction
2 Related Work
2.1 Data Sets for the CS TTS
2.2 Text Representation for CS TTS
3 Proposed Method
3.1 General Framework
3.2 CS Front-End
3.3 Synthesis Module

4 Data Description
5 Experiments
5.1 Input Representations
5.2 Experimental Setup
5.3 Experimental Results
6 Conclusion
References
Image Processing and Computer Vision
A Feature Fusion Network for Multi-modal Mesoscale Eddy Detection
1 Introduction
2 Related Work
2.1 Non-deep Learning Algorithms
2.2 Deep Learning Algorithms
3 Methodology
3.1 FusionNet
3.2 The Loss Function
4 Experiments
4.1 The Multi-modal Dataset
4.2 Experimental Results
5 Conclusion
References
A Hybrid Self-Attention Model for Pedestrians Detection
1 Introduction

2 Related Work
2.1 Pedestrian Detection
2.2 Attention Mechanism
3 Proposed Method
3.1 Revisiting the CSP Detector
3.2 Channel Attention
3.3 Spatial Attention
3.4 Hybrid Attention Fusion Strategy
4 Experiments
4.1 Dataset and Evaluation Metrics
4.2 Ablation Study
4.3 Comparison with State of the Arts
5 Conclusion
References
DF-PLSTM-FCN: A Method for Unmanned Driving Based on Dual-Fusions and Parallel LSTM-FCN
1 Introduction
2 Related Work
3 DF-PLSTM-FCN
3.1 Driving Model
3.2 Network Structure
3.3 Feature Fusion
3.4 Decision Fusion