Pan-African Artificial Intelligence and Smart Systems: First International Conference, PAAISS 2021, Windhoek, Namibia, September 6-8, 2021, ... and Telecommunications Engineering)

Preface
Organization
Contents
Deep Learning
A Critical Analysis of Deep Learning Architectures for Classifying Breast Cancer Using Histopathology Images
	1 Introduction
	2 Literature Review
		2.1 Problem Background
		2.2 Related Works
	3 Experimental Setup
	4 Method Comparison
		4.1 AlexNet Model
		4.2 DenseNet Model
		4.3 NASNet Mobile Model
	5 Results
	6 Recommendations
	7 Conclusion and Future Work
	References
A Patch-Based Convolutional Neural Network for Localized MRI Brain Segmentation
	1 Introduction
	2 MRI Segmmentation
		2.1 The Brain
		2.2 Magnetic Resonance Imaging (MRI)
		2.3 Segmentation
		2.4 Convolutional Neural Networks (CNN)
	3 Methodology
		3.1 Data Preprocessing
		3.2 Patch-Based CNN
		3.3 CNN Architecture
		3.4 The Proposed Algorithm
	4 Experimental Results and Discussions
	5 Conclusion and Recommendations
		5.1 Limitations
		5.2 Future Work
	References
Facial Recognition Through Localized Siamese Convolutional Neural Networks
	1 Introduction
	2 Literature Review
	3 Methods and Techniques
		3.1 Convolutional Neural Networks
		3.2 Siamese Architecture
	4 Loss Function
		4.1 Cascade Classifier Model
		4.2 Model Architecture
	5 Experimental Methodology
		5.1 Dataset
		5.2 Data Preprocessing
		5.3 Global Siamese CNN
		5.4 Patch-Based Siamese CNN
	6 Results
		6.1 Global Siamese CNN
		6.2 Patch-Specific CNNs
		6.3 Combination 1
		6.4 Combination 2
		6.5 Combination 3
	7 Discussion and Conclusion
	8 Future Work
	References
Classification and Pattern Recognition
Face Recognition in Databases of Images with Hidden Markov\'s Models
	1 Introduction
	2 Related Work
		2.1 Approaches Based on Statistics and Linear Algebra
		2.2 Face Recognition Using Neural Networks
		2.3 Stochastic Approaches in Face Recognition
	3 Comparison of Images Based on Hidden Markov Model
		3.1 Face Model Training Phase
		3.2 Calculation of the Normalized Similarity Rate Between Two Images
		3.3 Calculation of the Amplitude Coefficient
	4 Face Comparison
		4.1 Approach Based on Grid Partitioning
		4.2 Enhanced HMM: Our Approach Based on Clustering
	5 Experimental Results
		5.1 Synthesis of the Results of the Different Methods
		5.2 Commentary on the Results
	6 Conclusion and Perspectives
	References
Brain MRI Segmentation Using Autoencoders
	1 Introduction
	2 Literature Review
		2.1 Convolutional Neural Networks
		2.2 U-Net Models
		2.3 Autoencoders
		2.4 Review Summary
	3 Methodology
		3.1 Autoencoder Types
		3.2 Dataset
		3.3 Pre-processing
		3.4 Activation Function
		3.5 Loss Function
	4 Results and Discussion
		4.1 Evaluation Metrics
		4.2 Implementation
		4.3 Results
		4.4 Discussion
	5 Conclusion
	References
Effective Feature Selection for Improved Prediction of Heart Disease
	1 Introduction
	2 Related Works
	3 Methodology
		3.1 Hybrid SMOTE-ENN
		3.2 Recursive Feature Elimination
		3.3 Logistic Regression
		3.4 Decision Tree
		3.5 Random Forest
		3.6 Linear Discriminant Analysis
		3.7 Naïve Bayes
		3.8 Extreme Gradient Boosting
		3.9 The Architecture of the Proposed Heart Disease Prediction Model
	4 Dataset and Performance Metrics
	5 Results and Discussion
	6 Conclusion
	References
Convolutional Neural Network Feature Extraction for EEG Signal Classification
	1 Introduction
	2 Electroencephalograms
	3 Previous Work
	4 Methods an Techniques
		4.1 Convolutional Neural Networks
		4.2 Experimental Evaluation
	5 Results and Discussion
	6 Conclusion and Future Work
	References
Race Recognition Using Enhanced Local Binary Pattern
	1 Introduction
	2 Related Works
	3 Feature Extraction Methods
		3.1 Center Symmetric Local Binary Pattern (CSLBP)
	4 Proposed Method: Diagonal Symmetric Local Binary Pattern (DSLBP)
	5 Experiments
		5.1 Experiment 1
		5.2 Experiment 2
	6 Results and Discussions
		6.1 Experiment 1 Results
		6.2 Experiment 2 Results
		6.3 Robustness Testing of DSLBP
		6.4 Discussions
	7 Conclusion
	References
Detection and Classification of Coffee Plant Diseases by Image Processing and Machine Learning
	1 Introduction
	2 Releted Works
	3 Proposed Approach of Detection and Classification of Coffee Plant Diseases
		3.1 Preprocessing
		3.2 Segmentation
		3.3 Feature Extraction
		3.4 Feature Selection
		3.5 Classification
	4 Experiments
		4.1 Dataset Description
		4.2 Detection of Diseased Area
		4.3 Diseases Classification
	5 Results and Discussions
		5.1 Results
		5.2 Discussions
	6 Conclusion
	References
Plant Diseases Detection and Classification Using Transfer Learning
	1 Introduction
	2 Literature Review
		2.1 Plant Disease Classification using Deep Learning
		2.2 Transfer Learning Approach
	3 Methods and Techniques
		3.1 Dataset
		3.2 Pre-processing
		3.3 Deep Learning
		3.4 EfficientNets
		3.5 Softmax Classifier
		3.6 Transfer Learning
	4 Design And Implementation
		4.1 System Design and Experimental Setup
	5 Experimental Results and Discussion
		5.1 System Evaluation Matrix
		5.2 Results obtained
	6 Conclusion
	References
Neural Networks and Support Vector Machines
Hybridised Loss Functions for Improved Neural Network Generalisation
	1 Introduction
	2 Loss Functions in ANNs
		2.1 Sum Squared Error
		2.2 Cross Entropy
		2.3 The Effect of Loss Functions on Performance
	3 Hybrid Loss Functions
		3.1 Static Approach
		3.2 Adaptive Approach
		3.3 Reactive Approach
	4 Experimental Setup
		4.1 Datasets
		4.2 ANN Hyper-parameters
	5 Results
	6 Discussion
	7 Conclusion
	References
Diverging Hybrid and Deep Learning Models into Predicting Students’ Performance in Smart Learning Environments – A Review
	1 Introduction
		1.1 Contribution of the Study
	2 Method
		2.1 Search Strategy
		2.2 Inclusion and Exclusion Criteria
		2.3 Eligibility Criteria
	3 Results
		3.1 Predicting Students’ Academic Performance Using Deep Learning Techniques
		3.2 Deep Neural Networks for Predicting Students’ Academic Performance
		3.3 Hybrid Deep Learning Models for Predicting Students’ Performance
		3.4 Deep Long Short-term Memory Network for Predicting Students’ Performance
		3.5 Influential Factors for Predicting Students’ Academic Performance in Smart Learning Environments
	4 Conclusion
	References
Combining Multi-Layer Perceptron and Local Binary Patterns for Thermite Weld Defects Classification
	1 Introduction
	2 Materials and Methods
		2.1 Weld Joint Segmentation and Extraction
		2.2 Feature Extraction
		2.3 Feature Classification
	3 Experimental Results and Discussion
		3.1 Weld Joint Extraction
		3.2 Defect Classification
		3.3 Methods Comparison
	4 Conclusion
	References
Smart Systems
An Elliptic Curve Biometric Based User Authentication Protocol for Smart Homes Using Smartphone
	1 Introduction
	2 Related Work
	3 Proposed EC-B-ASH-S Scheme
		3.1 Initialization Phase
		3.2 Registration Phase
		3.3 Login and Authentication Phase
		3.4 Password/Biometric Change Phase
	4 Security Analysis and Performance Evaluation of the Proposed EC-B-ASH-S Scheme
		4.1 Formal Security Analysis Using BAN Logic
		4.2 Comparison of Computational Performance of Authentication Schemes
	5 Informal Security Analysis
		5.1 Confidentiality
		5.2 Masquerade Attack
		5.3 Replay Attack
		5.4 Denial of Service (DoS) and Distributed Denial of Service (DDoS) Attacks
		5.5 Perfect Forward Secrecy
		5.6 Man-in-the-Middle (MITM) Attack
		5.7 Password Guessing Attack
		5.8 Device Loss Attack
	6 Proof-of-Concept
		6.1 Registration Phase
		6.2 Login and Authentication Phase
	7 Conclusion
	References
Efficient Subchannel and Power Allocation in Multi-cell Indoor VLC Systems
	1 Introduction
	2 System Model
		2.1 VLC Network
		2.2 VLC Channel Model
	3 Sum-Rate Optimization
	4 Proposed Sum-Rate Optimization Solution
		4.1 Subchannel Allocation (SA) Procedure
		4.2 Power Allocation (PA) by the Quadratic Transform Approach
	5 Simulation Results and Discussions
	6 Conclusion
	References
Autonomic IoT: Towards Smart System Components with Cognitive IoT
	1 Introduction
	2 The Concept of IoST
	3 Device-Level Architecture for IoST
		3.1 Physical/Mechanical Plane
		3.2 Knowledge Plane
		3.3 Hardware Implementations
		3.4 Communication Technologies
	4 Case-Studies
		4.1 IoST in Energy
		4.2 IoST in Cyber-Physical Systems
	5 Conclusion
	References
Study of Customer Sentiment Towards Smart Lockers
	1 Introduction
	2 Literature Review
	3 Research Questions
	4 Survey Description and Distribution
	5 Methodology
	6 Results
		6.1 Respondent Breakdown
		6.2 Likert Scale Question Analysis
		6.3 Maximum Distance Analysis
		6.4 Short Answer Analysis
		6.5 Factors Effecting Use Results
	7 Discussion
	8 Limitations
	9 Conclusion
	References
Author Index