Artificial Neural Networks in Pattern Recognition: 10th IAPR TC3 Workshop, ANNPR 2022, Dubai, United Arab Emirates, November 24–26, 2022, Proceedings

Description

This book constitutes the refereed proceedings of the 10th IAPR TC3 International Workshop on Artificial Neural Networks in Pattern Recognition, ANNPR 2022, held in Dubai, UAE, in November 2022. The 16 revised full papers presented were carefully reviewed and selected from 24 submissions. The conference presents papers on subject such as pattern recognition and machine learning based on artificial neural networks.   Preface Organization Contents Learning Algorithms and Architectures Graph Augmentation for Neural Networks Using Matching-Graphs 1 Introduction and Related Work 2 Theory and Basic Models 2.1 Graphs 2.2 Graph Edit Distance (GED) 2.3 Graph Neural Networks (GNNs) 3 Augment Training Sets by Means of Matching-Graphs 4 Experimental Evaluation 4.1 Experimental Setup 4.2 Data Sets 4.3 Validation of Metaparameters 4.4 Test Results and Discussion 5 Conclusion and Future Work References A Novel Representation of Graphical Patterns for Graph Convolution Networks 1 Introduction 2 Related Work 3 The GrapHisto 4 Preliminary Experimental Evaluation 5 Conclusion References Minimizing Cross Intersections in Graph Drawing via Linear Splines 1 Introduction 2 Related Work 3 Method 3.1 Learning Non-differentiable Aesthetic Criteria: The Neural Aesthete 3.2 Employing Splines to Improve Graph Readability 3.3 Edge Crossing Optimization with Splines 3.4 Stress Optimization with Splines 4 Experiments 5 Conclusion References Multi-stage Bias Mitigation for Individual Fairness in Algorithmic Decisions 1 Introduction 2 Background and Related Work 2.1 Statistical Definitions of Fairness 2.2 Definitions of Individual Fairness 3 Multi-stage Individual Fairness 3.1 Notations 3.2 Transformed Representation Learning 3.3 Similarity Measure 3.4 Fairness Measure 3.5 Optimisation 4 Data and Experiment 4.1 Datasets 4.2 Evaluation Measures 4.3 Experimental Results 5 Conclusion References Do Minimal Complexity Least Squares Support Vector Machines Work? 1 Introduction 2 Minimal Complexity Least Squares Support Vector Machines 2.1 Architecture 2.2 Solving Subproblem 1 2.3 Solving Subproblem 2 2.4 Training Procedure 3 Performance Evaluation 4 Conclusions References A Review of Capsule Networks in Medical Image Analysis 1 Introduction 2 Background on Capsule Networks 2.1 Limitations of CNNs 2.2 Advantages of Capsule Networks 2.3 Capsule Network Architecture 3 Applications of Capsule Networks on Medical Images 3.1 Brain Injuries and Tumours 3.2 Ophthalmology 3.3 Cardiac Diseases 3.4 Pulmonary Diseases 4 Discussion 5 Conclusions and Recommendations for Future Work References Introducing an Atypical Loss: A Perceptual Metric Learning for Image Pairing 1 Introduction 2 Related Work 3 Learning Atypical Perceptual Similarity 3.1 The Baseline Triplet-Network 3.2 The Atypical Perceptual Similarity 4 Experimentation 4.1 The TTL Benchmark 4.2 Evaluation 5 Conclusion References Applications Wavelet Scattering Transform Depth Benefit, An Application for Speaker Identification 1 Introduction 2 Related Work 2.1 CNN-raw System 2.2 SincNet 2.3 HWSTCNN 3 Experimental Setup 3.1 Speaker Identification Text-Independent 3.2 Speaker Identification Text-Dependent 4 Results and Discussion 4.1 Speaker Identification Text-Independent 4.2 Speaker Identification Text-Dependent 5 Conclusion References Sequence-to-Sequence CNN-BiLSTM Based Glottal Closure Instant Detection from Raw Speech 1 Introduction 2 Data Description 2.1 Speech Material 2.2 GCI Detection Measures 3 Models 3.1 Baseline CNN-Based GCI Detection System 3.2 Recurrent Neural Network-Based GCI Detection 3.3 CNN-BiLSTM GCI Detection 4 Results 4.1 Comparison of Proposed Models 4.2 Comparison of Different GCI Detection Models 5 Conclusions References Mono vs Multilingual BERT for Hate Speech Detection and Text Classification: A Case Study in Marathi 1 Introduction 2 Related Work 3 Datasets 4 Experiments 4.1 Transformer Models 4.2 Evaluation Results 5 Conclusion References Transformer-Encoder Generated Context-Aware Embeddings for Spell Correction 1 Introduction 2 Related Work 2.1 Deep Learning Based Approaches to Spell Correction 3 Proposed Method 3.1 Model Architecture 3.2 Model Training and Triplet Loss 4 Experiment and Results 4.1 Dataset 4.2 Training, Evaluation and Baselines 4.3 Results 5 Conclusion and Future Work References Assessment of Pharmaceutical Patent Novelty with Siamese Neural Networks 1 Introduction 2 Related Work 2.1 Patent Content Analysis 2.2 Patent Relationships 2.3 Non-textual Analysis 3 Proposed Method 3.1 Data 3.2 Pipeline for Patent Document Processing 3.3 Creating Word Embeddings 3.4 Siamese Deep Neural Network Model 4 Results 4.1 Experimental Setup 4.2 Sentence and Document Embeddings Evaluation 4.3 Similarity Detection Model Evaluation 4.4 Ablation Studies 5 Discussion and Conclusion References White Blood Cell Classification of Porcine Blood Smear Images 1 Introduction 2 Methodology 2.1 Dataset 2.2 Model Implementation 2.3 Performance Evaluation 3 Results and Discussion 4 Conclusion References Medical Deepfake Detection using 3-Dimensional Neural Learning 1 Introduction 2 Dataset 3 Proposed Methodology 3.1 Detection Using Machine Learning 3.2 Detection Using 3DCNN 4 Results and Discussion 4.1 Experimental Results 5 Conclusion References A Study on the Autonomous Detection of Impact Craters 1 Introduction 2 Background 3 Dataset 4 Experimental Setup 4.1 Optimization Functions 4.2 Training Strategy 5 Results 6 Conclusion and Future Work References Utilization of Vision Transformer for Classification and Ranking of Video Distortions 1 Introduction 2 Materials and Methods 2.1 Dataset Overview 2.2 Vision Transformer – The Proposed Method 3 Results and Discussion 3.1 Experimental Setup 3.2 Experimental Results 4 Conclusion and Future Work References Author Index