18th International Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2023): Salamanca, Spain, September 5–7, ...

Preface
Organization
Contents
Special Session 2: Technological Foundations and Advanced Applications of Drone Systems
Level 3 Data Fusion
	1 Ontological Foundations
	2 Level 3 Data Fusion
	3 DNN Implementation
		3.1 CoA Recognition
		3.2 Event Prediction
		3.3 Forensic Assessment
	4 Modeling Courses of Action
		4.1 Probability of Action
		4.2 CoA Utility Modeling
		4.3 Evaluating Response Effectiveness
	References
Image Classification Using Contrastive Language-Image Pre-training: Application to Aerial Views of Power Line Infrastructures
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 Data Selection and Processing
		3.2 Data Labeling and Caption Generation
		3.3 CLIP Classifier Selection Methodology
	4 Results and Discussion
		4.1 Zero-Shot vs. Few-Shot Classifier
		4.2 CLIP Fine-Tuning
		4.3 Final model
	5 Conclusions
	References
A Realistic UAS Traffic Generation Tool to Evaluate and Optimize U-Space Airspace Capacity
	1 Introduction
	2 Realistic UAS Traffic Generation and Specification
		2.1 Identification of Common UAS Patterns
		2.2 Traffic Generation Module
		2.3 Traffic Specification Module
	3 Validation of the Traffic Generation and Specification Modules
	4 Next Steps: Towards a Separation Optimization and Evaluation Framework
	5 Conclusions
	References
UAV Airframe Classification Using Acceleration Spectrograms
	1 Introduction
	2 State of the Art
	3 Proposed System
		3.1 UAV Dataset and Spectrogram Generation
		3.2 Classification Algorithm
	4 Experiments and Results
	5 Conclusions
	References
Tuning Process Noise in INS/GNSS Fusion for Drone Navigation Based on Evolutionary Algorithms
	1 Introduction
	2 INS/GNSS
	3 Tuning Process
	4 Case Study
		4.1 Mission Problem and Simulation Configuration
		4.2 Filter Parameters
		4.3 Optimization Algorithms
	5 Results
		5.1 Results Comparison
	6 Conclusions
	References
Special Session 3: Soft Computing Methods in Manufacturing and Management Systems
Digital Twins of Production Systems Based on Discrete Simulation and Machine Learning Algorithms
	1 Introduction
	2 Reinforcement Learning
	3 A Digital Twin Based on Discrete-Event Simulation as a Reinforcement Learning Agent Environment
	4 Summary
	References
Edge Architecture for the Integration of Soft Models Based Industrial AI Control into Industry 4.0 Cyber-Physical Systems
	1 Introduction
	2 Related Work
	3 Architecture
	4 Validation
	5 Conclusions
	References
The Use of Line Simplification and Vibration Suppression Algorithms to Improve the Quality of Determining the Indoor Location in RTLSs
	1 Introduction
		1.1 Methods and Technologies for Determining Indoor Location
	2 Decawave as an Example of RTLS Based on UWB Technology
	3 Algorithms for Improving RTLS Data
		3.1 Polyline Simplification Algorithms
		3.2 Location Instability Suppression Algorithm (LISA)
	4 Research on the Effectiveness of Algorithms
		4.1 Discussion of Results
	5 Summary
	References
Possibilities of Decision Support in Organizing Production Processes
	1 Introduction
	2 Industry 4.0
		2.1 Technologies of the Industry 4.0
	3 Methodology
	4 Conclusion
	References
Special Session 4: Efficiency and Explainability in Machine Learning and Soft Computing
Efficient Short-Term Time Series Forecasting with Regression Trees
	1 Introduction
	2 Materials and Method
		2.1 Dataset
		2.2 Experimental Setup
		2.3 Evaluation Procedure
	3 Results
	4 Conclusions
	References
Generating Synthetic Fetal Cardiotocography Data with Conditional Generative Adversarial Networks
	1 Introduction
	2 Methodology
		2.1 Conditional Generative Adversarial Networks
		2.2 Classifiers
	3 Experiments
		3.1 Dataset
		3.2 CGAN Parameter Tuning
		3.3 Classifiers Hyperparameter and Parameter Tuning
	4 Results and Discussion
	5 Conclusions
	References
Olive Oil Fly Population Pest Forecasting Using Explainable Deep Learning
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 ADL
		3.2 Benchmark Algorithms
		3.3 Explainability
	4 Experimentation and Results
		4.1 Input Data
		4.2 Results and Discussion
		4.3 Explainability
	5 Conclusions and Future Work
	References
Explaining Learned Patterns in Deep Learning by Association Rules Mining
	1 Introduction
	2 Related Works
	3 Methodology
		3.1 Preprocessing
		3.2 Rules Mining
		3.3 Calculate Metrics
		3.4 Classify
	4 Results and Discussion
		4.1 Experimental Setting
		4.2 Metrics
		4.3 Results
	5 Conclusions
	References
Special Session 5: Machine Learning and Computer Vision in Industry 4.0
A Deep Learning Ensemble for Ultrasonic Weld Quality Control
	1 Introduction
	2 Technology
		2.1 Manufacturing Process
		2.2 Deep Learning Models
	3 Deep Learning Ensemble
	4 Experimental Results
	5 Conclusion
	References
Indoor Scenes Video Captioning
	1 Introduction
	2 Related Works
		2.1 Video Captioning
		2.2 Indoor Scene Captioning
	3 Methodology
	4 Experiments
		4.1 Charades Dataset
		4.2 Postprocessing
		4.3 Setup
		4.4 Results
	5 Conclusion
	References
A Multimodal Dataset to Create Manufacturing Digital Twins
	1 Introduction
	2 Related Work
		2.1 Pose Estimation for Action Recognition
	3 Experimental Setup and Data Acquisition
	4 Dataset Discussion
		4.1 Dataset Top View
		4.2 Dataset Side View
		4.3 Dataset Front View
	5 Datarecords
	6 Conclusions and Future Work
	References
A Modified Loss Function Approach for Instance Segmentation Improvement and Application in Fish Markets
	1 Introduction
	2 Dataset
	3 Proposed Method
	4 Experiments and Results
		4.1 Experimental Setup
		4.2 Results
	5 Conclusions
	References
Parallel Processing Applied to Object Detection with a Jetson TX2 Embedded System
	1 Introduction
	2 Methodology
	3 System Architecture
		3.1 Software Architecture
		3.2 Hardware Architecture
	4 Experimental Results
	5 Conclusion
	References
Deep Learning-Based Emotion Detection in Aphasia Patients
	1 Introduction
	2 Related Work
	3 Dataset
	4 Approach
	5 Evaluation
	6 Conclusion
	References
Defect Detection in Batavia Woven Fabrics by Means of Convolutional Neural Networks
	1 Introduction and Previous Work
	2 Related Work
	3 Case Study
	4 Methods and Experimentation
		4.1 Methods
		4.2 Experimentation
	5 Results
	6 Conclusions
	References
An Image Mosaicing-Based Method for Bird Identification on Edge Computing Devices
	1 Introduction
	2 Image Mosaicing-Based Method
	3 Experiments and Results
		3.1 Results
		3.2 Analysis of the Results
	4 Conclusion
	References
HoloDemtect: A Mixed Reality Framework for Cognitive Stimulation Through Interaction with Objects
	1 Introduction
	2 Related Works
	3 HoloLens 2 Application
		3.1 HoloLens 2 API
		3.2 Implementation Details
		3.3 Data Collection
	4 Evaluation of the Proposal
		4.1 Qualitative Analysis
		4.2 Quantitative Analysis
	5 Conclusions
	References
Accurate Estimation of Parametric Models of the Human Body from 3D Point Clouds
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 Obtaining and Preprocessing of the 3D Model.
		3.2 Estimation of an Intermediate Template Using BPS Neural Network
		3.3 First Minimization: BPS to SMPL
		3.4 Second Minimization: 3D Scan to SMPL
	4 Experiments
		4.1 Datasets and Evaluation Metrics
		4.2 Results
	5 Conclusions
	References
Lightweight Cosmetic Contact Lens Detection System for Iris Recognition at a Distance
	1 Introduction
	2 Related Work
	3 Overview of the IAAD Framework
	4 The Approach for Cosmetic Contact Lens Detection
		4.1 BSIF Encoding of an Iris Pattern
		4.2 Building the Ensemble of Classifiers
	5 Results
		5.1 Cross-Dataset Testing
	6 Conclusions and Future Work
	References
Vehicle Warning System Based on Road Curvature Effect Using CNN and LSTM Neural Networks
	1 Introduction
	2 Road Curvature-Based Dynamics of the Vehicle
	3 Risky Maneuvers Identification by CNN and LSTM Models
		3.1 Model’s Selection of Variables
		3.2 Deep Convolutional Neural Network Model
	4 Results and Discussion
	5 Conclusions and Future Works
	References
Special Session 6: Genetic and Evolutionary Computation in Real World and Industry
Enhancing Time Series Anomaly Detection Using Discretization and Word Embeddings
	1 Introduction
	2 Experimental Study
		2.1 Problem Formulation
		2.2 Data Preprocessing
		2.3 Model Architecture
		2.4 Datasets
	3 Results
	4 Conclusions and Future Work
	References
Multi-objective Optimization for Multi-Robot Path Planning on Warehouse Environments
	1 Introduction
	2 Non-Dominated Genetic Algorithm Approach
		2.1 Route Generation
		2.2 Initial Population
		2.3 Crossover
		2.4 Mutation
	3 Experimentation Setup
	4 Results and Discussion
	5 Conclusion and Future Work
	References
On the Prediction of Anomalous Contaminant Diffusion
	1 Introduction
	2 Bevilacqua-Galeão (BG) Model and Numerical Solution
		2.1 BG Model
		2.2 Numerical Solution
		2.3 Differential Evolution (DE) Method
	3 Inverse Problem Formulation
	4 Results
		4.1 Direct Problem and Case of Study
		4.2 Parameters Estimation
		4.3 Prediction of Concentration
	5 Conclusions and Future Work
	References
Keeping Safe Distance from Obstacles for Autonomous Vehicles by Genetic Algorithms
	1 Introduction
	2 Optimization Methodology
		2.1 Occupancy Map
		2.2 Distance to Obstacles Calculation
		2.3 Genetic Algorithm Fitness Function
	3 Simulation Results
	4 Conclusions and Future Works
	References
An Approach of Optimisation in Last Mile Delivery
	1 Introduction
	2 Related Work
	3 Modelling the Last Mile Delivery
		3.1 Parcel Delivery Models
		3.2 Mainframe of the Last Mile Delivery Algorithm
		3.3 Dataset
	4 Experimental Results
	5 Conclusion and Future Work
	References
Special Session 7: Soft Computing and Hard Computing for a Data Science Process Model
A Preliminary Study of MLSE/ACE-III Stages for Primary Progressive Aphasia Automatic Identification Using Speech Features
	1 Introduction
	2 Proposed Methodology
		2.1 Design of the Assessment Tool
		2.2 Intelligent Data Analisys
	3 Numerical Results
		3.1 Material and Methods
		3.2 Results
		3.3 Features Computing, Framing and Datasets Naming
		3.4 Classification Models Training, Results Discussion and Tasks Ranking
	4 Conclusions and Future Work
	References
Comparison of LSTM, GRU and Transformer Neural Network Architecture for Prediction of Wind Turbine Variables
	1 Introduction
	2 LSTM, GRU and Transformers Models and Case Study
		2.1 Transformer Model
		2.2 Case Study
	3 Results
		3.1 Data Pre-processing
		3.2 Training
		3.3 Prediction and Performance
	4 Conclusions and Future Works
	References
The Impact of Data Normalization on the Accuracy of Machine Learning Algorithms: A Comparative Analysis
	1 Introduction
	2 Background
	3 Experiments
	4 Results
		4.1 ML Algorithms and Normalization: A Graphical Approach
	5 Conclusions
	References
Adaptive Optics Correction Using Recurrent Neural Networks for Wavefront Prediction
	1 Introduction
	2 Adaptive Optics
	3 Neural Networks
		3.1 Long Short-Term Memory Neural Networks
	4 Experiment Description
		4.1 Training Data
		4.2 Network Architecture
	5 Experiments
		5.1 Training Performance by Varying the Number of Input Frames
		5.2 Training Performance by Varying the the Frequency of Slopes Attainment
	6 Results
		6.1 Variable Input Frames
		6.2 Variable Frequency
	7 Conclusion
	References
Author Index