Pattern Recognition and Artificial Intelligence: 4th Mediterranean Conference, MedPRAI 2020, Hammamet, Tunisia, December 20–22, 2020, Proceedings (Communications in Computer and Information Science)

This book constitutes the refereed proceedings of the 4th Mediterranean Conference on Pattern Recognition and Artificial Intelligence, MedPRAI 2020, held in Hammamet, Tunisia, in December 2020.​ Due to the COVID-19 pandemic the conference was held online. The 24 revised papers presented were thoroughly reviewed and selected from 72 submissions. The papers are covering the topics of recent advancements in different areas of pattern recognition and artificial intelligence, such as statistical, structural and syntactic pattern recognition, machine learning, data mining, neural networks, computer vision, multimedia systems, information retrieval, etc. Preface Organization Contents Computer Vision and Image Processing Fine-Tuning a Pre-trained CAE for Deep One Class Anomaly Detection in Video Footage 1 Introduction 2 Related 3 Proposed Method 3.1 Architecture 3.2 Training 3.3 Testing 4 Experimental Results 5 Conclusion References Plant Disease Recognition Using Optimized Deep Convolutional Neural Networks 1 Introduction 2 Related Works 3 The Proposed Framework 3.1 Dataset 3.2 The Proposed Method 3.3 EfficientNet Architecture 3.4 MobileNet V2 - Lite Version 4 Conclusion References APS: A Large-Scale Multi-modal Indoor Camera Positioning System 1 Introduction 2 Related Works 3 The Proposed Architecture 3.1 Created Dataset 3.2 Training of the Model 4 Experimental Analysis and Results 4.1 Testing of the Model 5 Conclusion References Improved Bilinear Model for Facial Expression Recognition 1 Introduction 2 Related Work 3 Approach 3.1 Bilinear CNN Models 3.2 Improved Bilinear Pooling 4 Experiments 4.1 Datasets 4.2 Model Architecture and Training Process 4.3 Ablation Study 4.4 Comparison with the State-of-the-Art 5 Conclusions and Perspectives References End-to-End Spectral-Temporal Fusion Using Convolutional Neural Network 1 Introduction 2 Related Works 3 Background 3.1 Sentinel-2 Satellite 3.2 Super-Resolution CNN 4 The Spectral-Temporal Fusion Technique 4.1 Datasets 5 Experimental Results and Discussion 5.1 Root Mean Square Error 5.2 Spectral Angle Mapper 5.3 Structural Similarity 6 Conclusion References Fusing Local and Global Features for Person Re-identification Using Multi-stream Deep Neural Networks 1 Introduction 2 Overview of the Proposed Method 2.1 Segmentation Step 2.2 Multi-stream Feature Extraction Method 2.3 Similarity Scores Fusion 3 Experiments 3.1 Datasets 3.2 Implementation 3.3 Results 4 Conclusion References Analysis of Histogram of Oriented Gradients on Gait Recognition 1 Introduction 2 Methodology 2.1 Gait Energy Image (GEI) 2.2 Gait Entropy Image (GEnI) 2.3 Gait Gaussian Image (GGI) 2.4 Gait Gaussian Entropy Image (GGEnI) 2.5 Gradient Histogram Gait Image (GHGI) 3 Evaluation 3.1 HoG Parameters 3.2 Testing Gait Representations 4 Conclusion References Possibilistic Classifier Combination for Person Re-identification 1 Introduction 2 Overview of the Proposed Method 2.1 Construction of Possibility Distributions 2.2 Aggregation of Possibility Distributions 3 Experiments 4 Conclusion References Implementation of Open-Source Image Analysis Techniques in Commercial Quality Control Systems 1 Introduction 2 Methodology 3 Evaluation 3.1 Testing Setup and Aims 3.2 Influence of Magnitude of ORB Feature Generation 3.3 Influence of Ratio Test Value 3.4 Influence of Offset 3.5 Influence of Illumination 3.6 Binary Testing 3.7 Processing Speed 4 Conclusion References Ensemble Classification Using Entropy-Based Features for MRI Tissue Segmentation 1 Introduction 2 Related Work 3 Pattern-Based Features for MRI Data Classification 3.1 AdaBoost for MRI Data Classification 3.2 MRI Data 3.3 Energy Coding-Based Classification 4 Experimentation and Evaluation 4.1 Parameter Selection 4.2 Performance Evaluation 4.3 Experimental Results 4.4 Results Analysis and Discussion 5 Conclusion References From a Textual Narrative to a Visual Story 1 Introduction 2 Related Work 3 Framework 4 Text Processing 5 Object Positioning 5.1 Bounding Box 5.2 Spatial Relationships 6 Scene Generation 7 Conclusion and Future Work References Document and Media Analysis Personality Traits Identification Through Handwriting Analysis 1 Introduction 2 Literature Review 3 Proposed Methodology 3.1 The Approach Overview 3.2 Dataset 3.3 Noise Removal 3.4 Feature Extraction 3.5 Classification 4 Experimental Evaluation 4.1 Methodology Principle 4.2 Analysis and Evaluation 5 Conclusion References Efficient Service Selection in Multimedia Documents Adaptation Processes 1 Introduction 2 Multimedia Documents Adaptation: State-of-the-Art 2.1 Multimedia Documents Adaptation in Pervasive Systems 2.2 Classification of Multimedia Adaptation Approaches 2.3 Adaptation Services Selection Issues 3 Contributions 3.1 Context and Motivations 3.2 Problem Statement and Modeling 3.3 Solving Methods 3.4 Synthetic Methods to Simulate Adaptation Service QoS Values 3.5 Finding Approximate Solutions Using Evolutionary Algorithms 4 Experiments, Numerical Results and Discussion 5 Conclusion References Offline Writer Identification Based on CLBP and VLBP 1 Introduction 2 Related Work 3 Feature Extraction 3.1 Complete Local Binary Patterns (CLBP) 3.2 Local Binary Pattern Variance (LBPV) 4 Decision Strategy 5 Experimental Results 6 Conclusions and Future Works References Deep Learning Architecture for Off-Line Recognition of Handwritten Math Symbols 1 Introduction 2 Related Works 3 Proposed System 3.1 Data Generation and Tuning 3.2 Model Tuning 4 Experimental Results 4.1 CROHME Dataset 4.2 Experimental Setup 4.3 Results and Discussion 5 Conclusion and Future Work References A Conditional GAN Based Approach for Distorted Camera Captured Documents Recovery 1 Introduction 2 Proposed Method 2.1 Generator 2.2 Recognizer 2.3 Discriminator 3 Experiments and Results 3.1 Dataset and Training Details 3.2 Evaluated Approaches and Metrics 3.3 Results 4 Conclusion References Learning Features for Writer Identification from Handwriting on Papyri 1 Introduction 2 Related Work 3 Materials and Methods 3.1 Dataset 3.2 Pre-processing 3.3 Data Preparation Using Dense Sampling 3.4 Two-Step Fine Tuning of ConvNets 4 Experiments and Results 5 Conclusion References Offline Signature Verification Using Feature Learning and One-Class Classification 1 Introduction 2 Related Work 3 Datasets 4 Proposed Methodology 4.1 Feature Learning Using CNN 4.2 One-Class Classification Using SVM 5 Experiments and Results 6 Conclusion References Artificial Intelligence and Intelligent Systems Text Mining-Based Association Rule Mining for Incident Analysis: A Case Study of a Steel Plant in India 1 Introduction 2 Literature Review 2.1 Research Issues 2.2 Contributions of the Work 3 Materials and Methods 3.1 Variables and Data Types 3.2 Methods 4 Results and Discussions 4.1 Extraction of the Best Item-Set Rules from Textual Data 4.2 Extraction of the Best Item-Set Rules from Categorical Data 5 Conclusions References Mining Interesting Association Rules with a Modified Genetic Algorithm 1 Introduction 2 Association Rules Mining 2.1 Definitions 2.2 Association Rules Mining as an Optimization Problem 2.3 Redundant Rules 3 Related Works 4 The Proposed Modified Genetic Algorithm for ARM 4.1 Objective Function and Rule Representation 4.2 Algorithm and Operators 5 Experiments and Results 5.1 Experimental Environment and Datasets 5.2 A Comparative Study with Other Approaches 5.3 Study on the Effect of Pruning Redundancy 6 Conclusion References ASCII Embedding: An Efficient Deep Learning Method for Web Attacks Detection 1 Introduction 2 Related Works 3 ASCII Embedding Method 4 Performance Evaluation of ASCII Embedding 4.1 Dataset Pre-processing 4.2 CNN Hyper-Parameters and Evaluation Metrics 4.3 Experimental Study and Results 4.4 Comparative Study and Discussion 5 Conclusion References Conditional Graph Pattern Matching with a Basic Static Analysis 1 Introduction 2 Background 2.1 Data Graphs and Graph Patterns 2.2 Conventional Graph Pattern Matching Revisited 3 Conditional Graph Patterns 3.1 Definition of CGPs 3.2 Conditional Simulation 4 An Algorithm for CGPs Matching 4.1 Description and Complexity Analysis 4.2 Parallel Version 5 Static Analysis 5.1 Pattern-Only Matching 5.2 Satisfiability of CGPs 5.3 Containment of CGPs 6 Experimental Study 7 Conclusion References An Effective Framework for Secure and Reliable Biometric Systems Based on Chaotic Maps 1 Introduction 2 System Description 2.1 Training Phase 2.2 Feature Extraction 2.3 Classification Phase 3 Experimental Results an Discussion 3.1 System Accuracy Analysis 3.2 System Security Analysis 4 Conclusion and Further Works References Learning Word Representations for Tunisian Sentiment Analysis 1 Introduction 2 TUNIZI 3 Tunisian Sentiment Analysis 4 Proposed Approach 4.1 Initial Representations 4.2 Classifiers 5 Results and Discussion 5.1 TUNIZI Dataset Results 5.2 TSAC-TUNIZI Dataset Results 5.3 Discussion 6 Conclusion and Future Work References Correction to: Pattern Recognition and Artificial Intelligence Correction to: C. Djeddi et al. (Eds.): Pattern Recognition and Artificial Intelligence, CCIS 1322, https://doi.org/10.1007/978-3-030-71804-6 Author Index