Artificial Neural Networks and Machine Learning – ICANN 2021: 30th International Conference on Artificial Neural Networks, Bratislava, Slovakia, September 14–17, 2021, Proceedings, Part III

Preface
Organization
Contents – Part III
Generative Neural Networks
Binding and Perspective Taking as Inference in a Generative Neural Network Model
	1 Introduction
	2 Proposed Model
		2.1 Sub-modal Population Encoding
		2.2 Gestalt Perception and Feature Binding
		2.3 Perspective Taking
	3 Experimental Results
		3.1 VAE Prediction Error
		3.2 Adaptation of Feature Binding
		3.3 Adaptation of Perspective Taking
	4 Summary and Conclusion
	References
Advances in Password Recovery Using Generative Deep Learning Techniques
	1 Introduction and Motivation
	2 Related Work
	3 Models
	4 Results
		4.1 Data
		4.2 Experimental Setup
		4.3 Analysis of Generated Passwords
		4.4 Password Guessing Performance
		4.5 Comparison to Established Methods
		4.6 Operations in Latent Space
	5 Conclusion and Future Work
	References
Dilated Residual Aggregation Network for Text-Guided Image Manipulation
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 Overview of DRA
		3.2 Encoder of DRA
		3.3 Decoder of DRA
		3.4 Objective Function of DRA
	4 Experiments
		4.1 Datasets
		4.2 Baselines
		4.3 Implementation Details
		4.4 Results
		4.5 Ablation Studies
		4.6 Failure Examples
	5 Conclusions
	References
Denoising AutoEncoder Based Delete and Generate Approach for Text Style Transfer
	1 Introduction
	2 Related Work
	3 Approach
		3.1 Delete
		3.2 Generate
	4 Experiments
		4.1 Datasets
		4.2 Models for Comparisons
		4.3 Evaluation Metrics
		4.4 Experiment Details
		4.5 Result Analysis
	5 Conclusion
	References
GUIS2Code: A Computer Vision Tool to Generate Code Automatically from Graphical User Interface Sketches
	1 Introduction
	2 Related Work
		2.1 Object Detection and Text Recognition
		2.2 GUI Code Generation
	3 Approach Description
		3.1 Overall Architecture
		3.2 UI Component Detection
		3.3 Text Recognition
		3.4 Code Parser
	4 Experimental Results and Analysis
		4.1 Dataset and Experiment Setup
		4.2 The Ablation Study and Evaluate
		4.3 Comparison with Other Approaches
	5 Conclusion
	References
Generating Math Word Problems from Equations with Topic Consistency Maintaining and Commonsense Enforcement
	1 Introduction
	2 Task Definition
	3 Model
		3.1 Variational Encoder-Decoder Module
		3.2 Enhancing Equation Encoder by Variational Autoencoder
		3.3 Topic Selection and Controlling
		3.4 Commonsense Enforcement
		3.5 Training Objective
	4 Experiments
		4.1 Datasets
		4.2 Motivation of Creating New Dataset
		4.3 Model Settings
		4.4 Automatic Evaluation
		4.5 Human Evaluation
		4.6 Case Study
	5 Conclusion
	References
Generative Properties of Universal Bidirectional Activation-Based Learning
	1 Towards More Brain-Like Learning
	2 UBAL Model
	3 Classification and Generative Properties
	References
Graph Neural Networks I
Joint Graph Contextualized Network for Sequential Recommendation
	1 Introduction
	2 Methods
		2.1 Graph Construction and Embedding
		2.2 Current Interest Module
		2.3 Global Preference Module
		2.4 Prediction Layer and Model Training
	3 Experimental Setup
	4 Results and Analysis
	5 Conclusion
	References
Relevance-Aware Q-matrix Calibration for Knowledge Tracing
	1 Introduction
	2 Related Work
	3 Preliminaries
	4 Method
		4.1 QKT
		4.2 RAQC
		4.3 RAQC for Knowledge Tracing
	5 Experiments
		5.1 Datasets
		5.2 Experimental Setup
		5.3 Experimental Results on Two Tasks
	6 Conclusion
	References
LGACN: A Light Graph Adaptive Convolution Network for Collaborative Filtering
	1 Introduction
	2 Related Work
		2.1 Collaborative Filtering
		2.2 Graph Convolution Network
	3 Methodology
		3.1 Embedding Layer
		3.2 Adaptive Embedding Propagation Layers
		3.3 Attention-Based Layer Combination and Model Prediction
	4 Experiments
		4.1 Datasets and Evaluation Metrics
		4.2 Compared Methods
		4.3 Parameter Settings
		4.4 Performance Comparison(RQ1)
		4.5 Performance Comparison with LightGCN(RQ2)
		4.6 Hyper-parameter Studies and Ablation Experiments (RQ3 and RQ4)
	5 Conclusion
	References
HawkEye: Cross-Platform Malware Detection with Representation Learning on Graphs
	1 Introduction
	2 Related Work
	3 Motivation
		3.1 Cross-Platform Malware Detection
		3.2 Representation Learning Based Feature Engineering
	4 System Design
		4.1 Architecture
		4.2 Graph Generator
		4.3 Feature Embedding
		4.4 MLP-Based Malware Classifier
	5 Evaluation
		5.1 Dataset and Experimental Setup
		5.2 Power Law and Opcode Embedding
		5.3 Evaluation Tasks
		5.4 Hyper-parameters Selection
		5.5 Detection on Obfuscated Samples
	6 Conclusion
	References
An Empirical Study of the Expressiveness of Graph Kernels and Graph Neural Networks
	1 Introduction
	2 Related Work
	3 Comparing Graphs to Each Other
	4 Empirical Evaluation
		4.1 Dataset
		4.2 Selected Approaches
		4.3 Baselines
		4.4 Experimental Settings
		4.5 Results
	5 Conclusion
	References
Multi-resolution Graph Neural Networks for PDE Approximation
	1 Introduction
	2 Related Works
		2.1 Machine Learning for PDEs
		2.2 Graph Neural Networks
	3 Multi-grid Graph Neural Networks
		3.1 Graph Convolutional Neural Networks
		3.2 Multi-grid Approaches
	4 Experimental Conditions
	5 Experimental Results
		5.1 Fixed Domains
		5.2 Variable Domains
		5.3 Computational Costs
	6 Conclusion
	References
Link Prediction on Knowledge Graph by Rotation Embedding on the Hyperplane in the Complex Vector Space
	1 Introduction
	2 Related Work
	3 Background
		3.1 RotatE
		3.2 TransH
	4 Proposed RotatH Model
	5 Experiments
		5.1 Datasets
		5.2 Parameters
		5.3 Metrics
		5.4 Running Environment
		5.5 Results
	6 Conclusion
	References
Graph Neural Networks II
Contextualise Entities and Relations: An Interaction Method for Knowledge Graph Completion
	1 Introduction
	2 Related Works
	3 Methodology
		3.1 Interaction Mechanism
		3.2 Interaction Embedding Decoders
	4 Experiments
		4.1 Benchmark Dataset
		4.2 Evaluation Protocol
		4.3 Main Results
		4.4 Analysis and Discussion
	5 Conclusion
	References
Civil Unrest Event Forecasting Using Graphical and Sequential Neural Networks
	1 Introduction
	2 Related Work
		2.1 Event Database
		2.2 Civil Unrest Forecasting
		2.3 Graph Convolutional Networks
	3 Methodology
		3.1 Construction of the Event Graph
		3.2 Representation Learning
		3.3 Prediction
	4 Experiments
	5 Results and Analysis
		5.1 Results
		5.2 The Influence of Lead Time
		5.3 The Influence of External Events
	6 Conclusions
	References
Parameterized Hypercomplex Graph Neural Networks for Graph Classification
	1 Introduction
	2 Related Work
	3 Hypercomplex Neural Networks
		3.1 Parameterized Hypercomplex Layer
	4 Hypercomplex Graph Neural Network
		4.1 Initialization of Linear Independent Contributions
		4.2 Hypercomplex Graph Neural Network
	5 Experiments
	6 Conclusion
	References
Feature Interaction Based Graph Convolutional Networks for Image-Text Retrieval
	1 Introduction
	2 Related Work
	3 Feature Interaction Based Graph Convolutional Networks for Image-Text Retrieval
		3.1 Feature Interaction Based Fragment Affinity Representation
		3.2 Relationship-Enhanced Fragment Feature Calculation
		3.3 Fragment-Level Image-Text Alignment
		3.4 Datasets and Evaluation Metric
	4 Experiment
		4.1 Implementation Details
		4.2 Comparisons with the State-of-the-Art
		4.3 Ablation Study
		4.4 Visualization
	5 Conclusion
	References
Generalizing Message Passing Neural Networks to Heterophily Using Position Information
	1 Introduction
	2 Related Work
	3 Problem
		3.1 Indices of Heterophily
	4 Approach
		4.1 Anchor Selection
		4.2 Position Encoder
		4.3 Representation Updater
		4.4 Complexity Analysis
	5 Experiments
		5.1 Datasets
		5.2 Compared Methods
		5.3 Experimental Setup
		5.4 Experimental Results
		5.5 Visualization
		5.6 Time Cost Analysis
	6 Conclusion
	References
Local and Non-local Context Graph Convolutional Networks for Skeleton-Based Action Recognition
	1 Introduction
	2 Related Work
		2.1 Skeleton Based Action Recognition
		2.2 Embedding High-Level Information in Graphs
		2.3 Modeling Long-Range Dependence
	3 Method
		3.1 Overview
		3.2 Motion Enhanced Graph
		3.3 Local and Non-local Context Module
		3.4 Optimization Strategies
	4 Experiment
		4.1 Dataset
		4.2 Implementation Details
		4.3 Ablation Study
		4.4 Comparison with the State-of-the-Art
	5 Conclusion
	References
STGATP: A Spatio-Temporal Graph Attention Network for Long-Term Traffic Prediction
	1 Introduction
	2 Prior Work
	3 Methodology
		3.1 Problem Definition
		3.2 Framework
		3.3 Temporal Convolutional Block
		3.4 Spatial Convolutional Block
	4 Experiments
		4.1 Datasets
		4.2 Experimental Settings
		4.3 Experimental Results
		4.4 Supplementary Experiment
	5 Conclusions
	References
Hierarchical and Ensemble Models
Integrating N-Gram Features into Pre-trained Model: A Novel Ensemble Model for Multi-target Stance Detection
	1 Introduction
	2 Related Works
	3 Model
		3.1 PMINFM Layer
		3.2 BiLSTM Layer
		3.3 Output and Model Training
	4 Experiment
		4.1 Dataset and Evaluation Metrics
		4.2 Preprocessing and Experimental Settings
		4.3 Results
		4.4 Case Study
	5 Conclusion
	References
Hierarchical Ensemble for Multi-view Clustering
	1 Introduction
	2 The Hierarchical Ensemble Framework
		2.1 The Proposed Model
		2.2 The Numerical Scheme
	3 Experiments
		3.1 Implementation
		3.2 Results and Discussion
	4 Conclusion
	References
Structure-Aware Multi-scale Hierarchical Graph Convolutional Network for Skeleton Action Recognition
	1 Introduction
	2 Related Works
		2.1 Traditional Methods for Action Recognition
		2.2 Graph-Based Methods for Action Recognition
	3 Method
		3.1 Graph Convolutional Network
		3.2 Pipeline Overview
		3.3 SA-HGP Block
		3.4 MSF Module
	4 Experiments
		4.1 Datasets
		4.2 Implementation Details
		4.3 Comparisons with the State-of-the-Art Methods
		4.4 Ablation Study
	5 Conclusion
	References
Learning Hierarchical Reasoning for Text-Based Visual Question Answering
	1 Introduction
	2 Related Work
	3 Method
		3.1 Multi-modal Embedding Construction
		3.2 Object Score Generator
		3.3 Text Modality Updater
		3.4 Answer Predictor
		3.5 Loss Design
	4 Experiment
		4.1 Datasets and Protocols
		4.2 Implementation Details
		4.3 Quantitative Analysis
		4.4 Ablation Study
		4.5 Case Study and Visualization
	5 Conclusion
	References
Hierarchical Deep Gaussian Processes Latent Variable Model via Expectation Propagation
	1 Introduction
	2 Preliminaries
		2.1 GP Regression and Approximation
		2.2 GP Sparse Approximations
	3 Methods
		3.1 Stochastic Expectation Propagation
		3.2 DGP-LVM
	4 Hierarchical Interaction Model
	5 Results
	6 Conclusion
	References
Adaptive Consensus-Based Ensemble for Improved Deep Learning Inference Cost
	1 Introduction
	2 Background
	3 Adaptive Consensus-Based Ensemble and Experiments
		3.1 Performance of Standard Ensemble
		3.2 Performance of Adaptive Consensus-Based Ensemble
	4 Concluding Remarks and Future Work
	References
Human Pose Estimation
Multi-Branch Network for Small Human Pose Estimation
	1 Introduction
	2 Related Work
		2.1 Top-Down Approaches
		2.2 Refined Operation
	3 Approach
		3.1 Multi-Branch Expansion Module
		3.2 Multi-Branch Downsample Module
		3.3 Refine Module
	4 Experiments
		4.1 Dataset and Evaluation Metric
		4.2 Implementation Details
		4.3 Ablation Study
		4.4 Comparison with State-of-the-art Methods
	5 Conclusion
	References
PNO: Personalized Network Optimization for Human Pose and Shape Reconstruction
	1 Introduction
	2 Related Work
	3 Method
		3.1 Problem Formulation
		3.2 Personalized Network Optimization
		3.3 Geometry-Aware Temporal Constraints
	4 Experiments
		4.1 Comparing with the State-of-the-Arts Methods
		4.2 Ablation Study
		4.3 Qualitative Evaluation
	5 Conclusion
	References
JointPose: Jointly Optimizing Evolutionary Data Augmentation and Prediction Neural Network for 3D Human Pose Estimation
	1 Introduction
	2 Related Work
	3 Evolutionary Data Augmentation (EDA)
		3.1 3D Human Skeleton Representation
		3.2 Evolutionary Data Augmentation
	4 Joint Training of EDA and HPE
		4.1 Pre-training of HPE Model
		4.2 Strategy of Reward/Penalty Function
	5 Experiments
		5.1 Datasets, Evaluation Metrics and Implementation Details
		5.2 Comparison with State-of-the-Art Methods
		5.3 Visualization of the Training Status
	6 Conclusion
	References
DeepRehab: Real Time Pose Estimation on the Edge for Knee Injury Rehabilitation
	1 Introduction
	2 Related Works
	3 Methods
		3.1 Refined PoseNet Outputs with Filtering Methods
		3.2 DeepRehab for Pose Estimation on the Edge
	4 Experiments
		4.1 Filtering Configurations
		4.2 PoseNet and DeepRehab Configurations
	5 Results
		5.1 Refined PoseNet Outputs with Filtering Methods
		5.2 DeepRehab for Pose Estimation on the Edge
	6 Discussion and Summary
	References
Image Processing
Subspace Constraint for Single Image Super-Resolution
	1 Introduction
	2 Related Work
	3 The Proposed Method
		3.1 Wavelet Transform
		3.2 Design of the WaveLoss Function
		3.3 Network Architecture
	4 Experiment and Analysis
		4.1 Datasets and Metrics
		4.2 Effectiveness of Subspace Constraint
		4.3 Comparisons with the State-of-the-Art Methods
		4.4 Object Recognition Performance
	5 Conclusion
	References
Towards Fine-Grained Control over Latent Space for Unpaired Image-to-Image Translation
	1 Introduction
	2 Related Work
	3 Methods
		3.1 Preliminaries
		3.2 Framework Architecture
		3.3 Loss Functions
	4 Experiments
	5 Results
	6 Conclusions
	References
FMSNet: Underwater Image Restoration by Learning from a Synthesized Dataset
	1 Introduction
	2 Related Work
	3 Proposed Method
		3.1 Synthesizing Underwater Dataset
		3.2 Proposed FMSNet CNN Framework
	4 Experimental Evaluation and Discussion
		4.1 Datasets and Training Strategy
		4.2 Network Performance
		4.3 Evaluation on Synthesized Underwater Images
		4.4 Evaluation on Real-World Underwater Images
		4.5 Evaluation by the Subsequent Application
	5 Conclusion
	References
Towards Measuring Bias in Image Classification
	1 Introduction
	2 Related Work
	3 An Approach for Measuring Data Bias
		3.1 Dataset Generation
		3.2 Attribution Methods
		3.3 Metrics
	4 Experimental Results
		4.1 Models
		4.2 Measure Bias Based on Metrics
	5 Discussion and Conclusion
	References
Towards Image Retrieval with Noisy Labels via Non-deterministic Features
	1 Introduction
	2 Related Work
	3 Proposed Model
		3.1 Non-deterministic Features
		3.2 Model Structure
		3.3 Sampling Non-deterministic Features
		3.4 Relative Entropy Loss Function
	4 Experiments and Analysis
		4.1 Datasets
		4.2 Settings and Implementation Details
		4.3 Comparison with Deterministic Features
		4.4 Influence of
		4.5 Analysis on Feature Variance
		4.6 Comparison with Hard Mining
	5 Conclusion
	References
Image Segmentation
Improving Visual Question Answering by Semantic Segmentation
	1 Introduction
	2 Related Work
	3 Proposed Method
		3.1 Semantic Segmentation for the Visual Encoder
		3.2 Application to Existing Models
	4 Experiments
	5 Conclusion
	References
Weakly Supervised Semantic Segmentation with Patch-Based Metric Learning Enhancement
	1 Introduction
	2 Related Work
		2.1 Weakly Supervised Semantic Segmentation
		2.2 Metric Learning
	3 Weakly Supervised Semantic Segmentation with Patch-Based Metric Learning Enhancement
		3.1 Classification Task
		3.2 CAM-Based Object Proposal
		3.3 Patch-Based Metric Learning
		3.4 Joint Training Scheme
	4 Experiment
		4.1 Experimental Setting
		4.2 CAM Analysis
		4.3 State-of-the-Art Methods Comparison
		4.4 Ablation Study on Hard Sampling Strategy
		4.5 Computation Complexity
	5 Conclusion
	References
ComBiNet: Compact Convolutional Bayesian Neural Network for Image Segmentation
	1 Introduction
	2 Related Work
	3 ComBiNet
		3.1 Repeat Block
		3.2 Bayesian Inference
	4 Experiments
		4.1 CamVid
		4.2 Bacteria
		4.3 Discussion
	5 Conclusion
	References
Depth Mapping Hybrid Deep Learning Method for Optic Disc and Cup Segmentation on Stereoscopic Ocular Fundus
	1 Introduction
	2 Related Work
		2.1 Optic Disc and Cup Segmentation
		2.2 Retinal Vessel Segmentation
		2.3 Multi-view Stereo Depth
	3 Depth Mapping Hybrid Deep Learning Method for ODC Segmentation
		3.1 Vessel Removing and Depth Computation
		3.2 Fusion of ROI Features and Depth Mapping
		3.3 Loss Function
	4 Experiments
		4.1 Implementation
		4.2 Datasets and Networks
		4.3 Evaluation Metrics
		4.4 Results and Analysis
	5 Conclusion
	References
RATS: Robust Automated Tracking and Segmentation of Similar Instances
	1 Introduction
	2 Methods
		2.1 Pre-training
		2.2 Temporal Extension Using Tracktor
		2.3 Propagation and Error Detection
		2.4 Annotation Needs and Fine-Tuning
		2.5 Benchmark Measures
	3 Results
		3.1 Experimental Setup
		3.2 Database
		3.3 Temporal Extension
		3.4 Robustness
		3.5 Error Detection and Fine-Tuning
	4 Summary
	References
Knowledge Distillation
Data Diversification Revisited: Why Does It Work?
	1 Introduction
	2 Background
	3 How Does Data Diversification Benefit NMT Models?
		3.1 Comparative Analysis and Hypothesis
		3.2 Preliminary Study for Modality
		3.3 Metric for Data Modality Complexity
		3.4 Hypothesis Verification
	4 Does the Performance Improvement Come from Knowledge Distillation?
	5 How to Reduce the Computational Cost of Data Diversification?
		5.1 Adjustable Sampling
		5.2 Clean Tuning
		5.3 Experiments
	6 Conclusion
	References
A Generalized Meta-loss Function for Distillation Based Learning Using Privileged Information for Classification and Regression
	1 Introduction
	2 Methods
		2.1 Proposed Formulation
		2.2 Experiments
	3 Results and Discussion
		3.1 Synthetic Datasets
		3.2 MNIST Handwritten Digit Image Classification
		3.3 Protein Binding Affinity Prediction
	4 Conclusion
	References
Empirical Study of Data-Free Iterative Knowledge Distillation
	1 Introduction
	2 DF-IKD
		2.1 Refinement Operator
		2.2 Loss Functions
	3 Empirical Evaluation
		3.1 Aim
		3.2 Materials
		3.3 Method
		3.4 Results
	4 Related Work
	5 Conclusion
	References
Adversarial Variational Knowledge Distillation
	1 Introduction
	2 Related Works
		2.1 Knowledge Distillation
		2.2 Data-Free Knowledge Distillation
	3 Method
		3.1 Problem Formulation
		3.2 Variational Knowledge Distillation (VKD)
		3.3 Adversarial Variational Knowledge Distillation (AVKD)
		3.4 Algorithm
	4 Experiments
		4.1 Experiments Setup
		4.2 Implementation Details
		4.3 Results
	5 Conclusion
	References
Extract then Distill: Efficient and Effective Task-Agnostic BERT Distillation
	1 Introduction
	2 Preliminary
		2.1 Architecture of BERT
		2.2 Architecture of Transformer
		2.3 Overview of BERT Distillation
	3 Methodology
	4 Experiments
		4.1 Experimental Setup
		4.2 Main Results
	5 Ablation Study
		5.1 The Effect of Each Module
		5.2 The Strategy of Layer Selection
	6 Related Work
	7 Conclusion and Future Work
	References
Medical Image Processing
Semi-supervised Learning Based Right Ventricle Segmentation Using Deep Convolutional Boltzmann Machine Shape Model
	1 Introduction
	2 Convolutional Deep Boltzmann Machine
		2.1 Architecture of CDBM
		2.2 Training
	3 Semi-supervised Learning Network
		3.1 Network Training
	4 Experiments and Results
	5 Conclusion
	References
Improved U-Net for Plaque Segmentation of Intracoronary Optical Coherence Tomography Images
	1 Introduction
	2 Data and Method
		2.1 Data Acquisition and Processing
		2.2 Plaque Segmentation Method
	3 Experimental Results and Discussion
	4 Conclusion
	References
Approximated Masked Global Context Network for Skin Lesion Segmentation
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 Approximated Masked Global Context
		3.2 Context Modeling Module
		3.3 Channel Attention Module
		3.4 Hybrid Loss Function
	4 Experiments
		4.1 Datasets
		4.2 Implementation Details
		4.3 Ablation Studies of Different Modules
		4.4 Comparison to State-of-the-Art
	5 Conclusion
	References
DSNet: Dynamic Selection Network for Biomedical Image Segmentation
	1 Introduction
	2 Related Work
		2.1 Model Architectures for Medical Image Segmentation
		2.2 Receptive Field
	3 Methods
		3.1 Dynamic Selection Module
		3.2 Overall Architecture
		3.3 Loss Function
	4 Experiments
		4.1 Datasets
		4.2 Implementation Details
		4.3 Uterus Segmentation
		4.4 Gland Segmentation and Lung Segmentation
		4.5 Ablation Studies
	5 Conclusion
	References
Computational Approach to Identifying Contrast-Driven Retinal Ganglion Cells
	1 Introduction
	2 Methods
		2.1 Modelling Sub-receptive Fields
		2.2 Assessing Input-Output Relationship
		2.3 Modelling Retinal Behaviour
	3 Performance Evaluation
	4 Discussion
	5 Conclusion
	References
Radiological Identification of Hip Joint Centers from X-ray Images Using Fast Deep Stacked Network and Dynamic Registration Graph
	1 Introduction
	2 The Proposed Method
		2.1 Method Architecture
		2.2 Landmark Extraction
		2.3 Grey Relational Analysis
		2.4 Fast HJC Prediction Module
		2.5 Accurate HJC Positioning Module
	3 Experimental Results
		3.1 Data Processing and Analysis
		3.2 Experimental Settings
		3.3 Comparison with State-of-the-Art Methods
		3.4 Ablation Analysis
	4 Conclusion
	References
A Two-Branch Neural Network for Non-Small-Cell Lung Cancer Classification and Segmentation
	1 Introduction
	2 Related Work
		2.1 Semantic Segmentation
		2.2 Pathological Classfication
	3 Method
		3.1 Segmentation Branch
		3.2 Classification Branch
	4 Experiments
		4.1 Dataset
		4.2 Implementation Detail
		4.3 Results and Analysis
	5 Conclusion
	References
Uncertainty Quantification and Estimation in Medical Image Classification
	1 Introduction
	2 Background and Related Work
		2.1 Bayesian Probabilistic Modeling
		2.2 Monte-Carlo Dropout as a Bayesian Approximation in Neural Networks
		2.3 Deep Ensemble Approach
		2.4 Ensemble MC Dropout Approach
		2.5 Medical Pre-screening
	3 Datasets
	4 Experiments and Results
		4.1 Data Preprocessing
		4.2 Uncertainty Quantification on SARS-CoV2
		4.3 Uncertainty Quantification on BreaKHis
	5 Conclusion and Future Work
	References
Labeling Chest X-Ray Reports Using Deep Learning
	1 Introduction
	2 Related Work
	3 Dataset and Method
		3.1 Dataset
		3.2 Language Model
		3.3 Multi-label Classifier
	4 Experiment
	5 Evaluation
		5.1 Result
		5.2 Discussion and Future Work
		5.3 Conclusion
	References
Author Index