Intelligent Systems : 11th Brazilian Conference, BRACIS 2022, Campinas, Brazil, November 28 – December 1, 2022, Proceedings, Part I

The two-volume set LNAI 13653 and 13654 constitutes the refereed proceedings of the 11th Brazilian Conference on Intelligent Systems, BRACIS 2022, which took place in Campinas, Brazil, in November/December 2022. The 89 papers presented in the proceedings were carefully reviewed and selected from 225 submissions. The conference deals with theoretical aspects and applications of artificial and computational intelligence. Preface Organization Contents – Part I Contents – Part II Mortality Risk Evaluation: A Proposal for Intensive Care Units Patients Exploring Machine Learning Methods 1 Introduction 2 Related Works Exploring Machine Learning to Predict Mortality Risk in ICUs 3 Prediction of Mortality Risk in ICUs: Approach Design 3.1 Discussion of the Research Problem 3.2 Database and Study Population 3.3 Exploratory Data Analysis and Variable Selection 3.4 Data Preprocessing 3.5 Feature Construction and Data Normalization 4 Prediction of Mortality Risk in ICUs: Approach Evaluation 4.1 Tuning of the Best Performing Machine Learning Method 4.2 Performance Analysis of the Best Performance Method 5 Conclusions References Requirements Elicitation Techniques and Tools in the Context of Artificial Intelligence 1 Introduction 2 Background 2.1 Software Requirements 2.2 Ethical Requirements for AI 2.3 Related Works 3 Research Methodology 4 Survey Results and Discussion 4.1 Threats to Validity 5 Conclusions References An Efficient Drift Detection Module for Semi-supervised Data Classification in Non-stationary Environments 1 Introduction 2 Background 2.1 Flexible Confidence of a Classifier Semi-supervised Technique 2.2 Classifier Ensemble 2.3 Data Stream Classification 3 Related Work 4 The Proposed Approach 5 Experimental Methodology 6 Experimental Results 6.1 Batch Size Analysis 6.2 The Proposed Methods versus DyDaSL - FT 6.3 DyDaSL versus State-of-Art 7 Final Remarks References The Impact of State Representation on Approximate Q-Learning for a Selection Hyper-heuristic 1 Introduction 2 Reinforcement Learning 3 Selection Hyper-heuristic 4 Proposed Approach 4.1 State Module 4.2 Reward Module 4.3 Agent Module 5 Experimental Setup 6 Results and Discussion 6.1 Bin Packing 6.2 Flow Shop 6.3 MAX-SAT 6.4 Personnel Scheduling 6.5 Traveling Salesman Problem 6.6 Vehicle Routing Problem 6.7 Overall Comparison 7 Conclusion References A Network-Based Visual Analytics Approach for Performance Evaluation of Swarms of Robots in the Surveillance Task 1 Introduction 2 Related Work 2.1 Visualisation in Robotics 2.2 Visualisation of Temporal Networks 2.3 PheroCom Model 3 Visualisation Proposal 4 Case Study 4.1 Surveillance Network 4.2 Experiments 5 Limitations 6 Conclusion and Future Work References Ulysses-RFSQ: A Novel Method to Improve Legal Information Retrieval Based on Relevance Feedback 1 Introduction 2 Literature Review 2.1 Legal Information Retrieval 2.2 Relevance Feedback and Its Use for Similar Queries 3 Ulysses-RFSQ: Improving LIR With Relevance Feedback 3.1 Step 1: Ranking The Documents 3.2 Step 2: Selecting Similar Queries 3.3 Step 3: Updating The Ranking 3.4 Step 4: Acquiring The Relevance Feedback Information 4 Experimental Setup 4.1 Corpora 4.2 Pre-processing 4.3 BM25 Algorithms 4.4 Cut-off Parameter 4.5 IR Evaluation Measure 5 Results and Discussion 5.1 BM25 Algorithms Comparison 6 Conclusion References Adaptive Fast XGBoost for Regression 1 Introduction 2 Related Works 3 Proposal: Adaptive Fast XGBoost Regressor 3.1 Implementation 4 Results Assessment 4.1 Testing Methodology 4.2 Analysis of the Results 5 Conclusion and Future Work References The Effects of Under and Over Sampling in Exoplanet Transit Identification with Low Signal-to-Noise Ratio Data 1 Introduction 2 Related Work 3 Folding the Light Curve 4 Materials and Methods 5 Results 6 Discussion 7 Conclusions References Estimating Bone Mineral Density Based on Age, Sex, and Anthropometric Measurements 1 Introduction 2 Related Work 3 Methodology 3.1 Dataset 3.2 Learning Process 3.3 Evaluation 3.4 Interpretation 4 Results and Discussion 4.1 Cross-Validation Results 4.2 Interpretation 5 Conclusion References Feature Extraction for a Genetic Programming-Based Brain-Computer Interface 1 Introduction 2 Brain-Computer Interface and Post-stroke Motor Rehabilitation 3 Datasets 3.1 BCI Competition IV 2a 3.2 BCI Competition IV 2b 3.3 Our Dataset 4 Proposed Method 4.1 Initial Feature Extraction 4.2 Genetic Programming 5 Computational Experiments 5.1 Analysis of the Number of Electrodes 5.2 Analysis of Within and Cross-Session Training 5.3 Analysis Using Data Obtained with Low-Cost Equipment 6 Concluding Remarks and Future Work References Selecting Optimal Trace Clustering Pipelines with Meta-learning 1 Introduction 2 Related Work 3 Problem Statement 4 MtL-Based Solution for Trace Clustering 5 Experimental Setup 5.1 Event Logs and Featurization 5.2 Trace Encoding Techniques 5.3 Trace Clustering Algorithms 5.4 Ranking Metrics 5.5 Meta-model 6 Results and Discussion 6.1 Meta-learning Exploratory Analysis 6.2 Meta-model Performance 7 Conclusion 8 Limitations and Broader Impact Statement References Sequential Short-Text Classification from Multiple Textual Representations with Weak Supervision 1 Introduction 2 Related Work 3 Methods 3.1 Labeling Function 3.2 TD-BERT 4 Evaluation 4.1 Datasets 4.2 Pre-processing 4.3 Classification Models and Experimental Setup 4.4 Results and Discussion 5 Conclusion References Towards a Better Understanding of Heuristic Approaches Applied to the Biological Motif Discovery 1 Introduction 2 Literature Review 3 Problem Definition and Algorithms 3.1 VNS 3.2 EM 3.3 ILS 3.4 Constructive Procedure 4 Experiments 4.1 Datasets 4.2 Results and Discussion 4.3 Statistical Analysis 5 Conclusion References Mutation Rate Analysis Using a Self-Adaptive Genetic Algorithm on the OneMax Problem 1 Introduction 2 Background 2.1 Genetic Algorithm 2.2 Fuzzy Logic 3 Methodology 3.1 Test Environment 3.2 Problem Description 3.3 Genetic Algorithm 3.4 Application of Fuzzy Logic 3.5 Evaluation Metrics 4 Results and Discussions 5 Conclusion References Application of the Sugeno Integral in Fuzzy Rule-Based Classification 1 Introduction 2 Preliminary Concepts and the Sugeno-like Generalization 3 Application of the Sugeno Integral to Classification in FRBCS 3.1 The New Fuzzy Reasoning Method 3.2 Experimental Framework 4 Experimental Results 4.1 Statistical Analysis 5 Conclusion References Improving the FQF Distributional Reinforcement Learning Algorithm in MinAtar Environment 1 Introduction 2 Reinforcement Learning 3 Related Work 3.1 DQN 3.2 Prioritized Experience Replay 3.3 Rainbow 3.4 Distributional Reinforcement Learning and FQF 3.5 Munchausen R.L. 3.6 MinAtar 4 Methodology 5 Experiments and Analysis of Results 5.1 Hyperparameter Tuning 5.2 Main Results 6 Conclusions and Future Works References Glomerulosclerosis Identification Using a Modified Dense Convolutional Network 1 Introduction 2 Related Works 3 Materials and Methods 3.1 Proposed Methodology 3.2 Image Dataset 3.3 Pre-processing and Data Augmentation 3.4 Evaluated Convolutional Neural Networks 3.5 Transfer Learning and Fine-Tunning 3.6 Evaluation Metrics 4 Results and Discussion 5 Conclusion and Future Works References Diffusion-Based Approach to Style Modeling in Expressive TTS 1 Introduction 2 Related Works 3 Background 3.1 Denoising Diffusion Probabilistic Models 3.2 Shallow Diffusion Mechanism 4 Model 4.1 Model Architecture 4.2 Training and Inference 5 Experiments 5.1 Experimental Setup 5.2 Naturalness 5.3 Style Transfer 6 Discussion 7 Conclusion References Automatic Rule Generation for Cellular Automata Using Fuzzy Times Series Methods 1 Introduction 2 Background 2.1 Cellular Automata 2.2 Fuzzy Time Series 3 Related Work 4 Proposed Method 4.1 Training Procedure 4.2 Forecast Procedure 5 Computational Experiments 5.1 Dataset Description 5.2 CA-FTS Modeling 5.3 Discussion 6 Conclusions and Future Work References Explanation-by-Example Based on Item Response Theory 1 Introduction 2 Background 2.1 Explainable Artificial Intelligence - XAI 2.2 Item Response Theory - IRT 3 Methodology 3.1 ML and IRT 3.2 Evaluated Datasets 4 Results and Discussion 4.1 Datasets Through the Lens of IRT 4.2 Random Forest Through the Lens of IRT 5 Final Considerations References Short-and-Long-Term Impact of Initialization Functions in NeuroEvolution 1 Introduction 2 Related Work 3 Theoretical Foundation 3.1 CoDeepNEAT 3.2 Short, Medium and Long Term Analyses 3.3 Initialization and Activation Functions 4 Experiments and Discussion 5 Conclusion References Analysis of the Influence of the MVDR Filter Parameters on the Performance of SSVEP-Based BCI 1 Introduction 2 Methodology 2.1 Database Description 2.2 CAR 2.3 MVDR Filter 2.4 Feature Extraction 2.5 Linear Classifier 3 Results and Discussion 4 Conclusion References A Novel Multi-objective Decomposition Formulation for Per-Instance Configuration 1 Introduction 2 Related Works 3 MOAAC/D: Multi-objective Automated Algorithm Configuration Based on Problem Space Decomposition 3.1 Decomposing the Problem Space: From AAC Toward MOAAC/D 3.2 iMOEAD: Solving the Decomposed Problem 3.3 Decision Maker: Recommending Configurations 4 Experiments on Flowshop Problems 5 Results 6 Conclusion References Improving Group Search Optimization for Automatic Data Clustering Using Merge and Split Operators 1 Introduction 2 Group Search Optimization (GSO) 3 Proposed Approaches: MGSO, SGSO and MSGSO 4 Experimental Results 5 Conclusions References Leveraging Textual Descriptions for House Price Valuation 1 Introduction 2 Related Work 3 Methodology 3.1 Data Collection 3.2 Pre-processing and Extraction Methods 3.3 Learning Algorithms 3.4 Model Evaluation 3.5 Interpretability Setup 4 Results 4.1 Generalization Analysis 4.2 Qualitative Analysis 5 Conclusion References Measuring Ethics in AI with AI: A Methodology and Dataset Construction 1 Introduction 2 Background and Related Work 3 On Building a Dataset for Measuring Ethics in AI 3.1 Active Learning 3.2 Dataset Analysis 4 The Construction of the AI-Based AI-Index 4.1 A Logistic Regression Model 4.2 Re-Analysing the AI-Index 5 Limitations 6 Discussion References Time Robust Trees: Using Temporal Invariance to Improve Generalization 1 Introduction 2 Related Work 3 Learning Time Robust Trees 3.1 Motivational Example 3.2 Time Robust Forests 3.3 Synthetic Example 3.4 Hyper-parameter Optimization 4 Experiments 5 Discussion and Conclusion References Generating Diverse Clustering Datasets with Targeted Characteristics 1 Introduction 2 Instance Space Analysis 3 Meta-dataset and ISA Results 3.1 Clustering Datasets 3.2 Meta-features 3.3 Algorithm Portfolio and Footprints 3.4 Analysis of ISA Results 4 Generation of Artificial Problem Instances 4.1 Proposed Method 4.2 Generated Datasets and Discussion 5 Final Considerations References On AGM Belief Revision for Computational Tree Logic 1 Introduction 2 Preliminaries 3 Computational Tree Logic 4 Temporal Preference Logic 4.1 Belief Change in CTL 5 Conclusions References Hyperintensional Models and Belief Change 1 Introduction 2 Related Work 3 Preliminaries 4 Hyperintensional Belief Contractions 5 A Model of Hyperintensional Belief Change 6 Conclusions References A Multi-population Schema Designed for Biased Random-Key Genetic Algorithms on Continuous Optimisation Problems 1 Introduction 2 Background 2.1 Biased Random-Key Genetic Algorithms 2.2 Distributed Genetic Algorithms 3 Proposed Model 4 Design of Experiments 4.1 Parameter Settings 5 Results and Analysis 6 Conclusion and Future Works References Answering Questions About COVID-19 Vaccines Using ChatBot Technologies 1 Introduction 2 Background 2.1 Chatbot Technologies 2.2 COVID-19 Pandemic 3 A Chatbot Extended Architecture 3.1 Most Common Questions About COVID-19 Vaccines 3.2 NLU Training 4 Knowledge Engineering 4.1 Dialogue Strategies 5 Empirical Evaluation 5.1 Evaluating the NLU 5.2 Evaluating the Dialogue Strategies 6 Related Work 7 Conclusion References Analysis of Neutrality of AutoML Search Spaces with Local Optima Networks 1 Introduction 2 Related Work 3 Problem Definition and Fitness Landscape 4 Local Optimal Networks 4.1 Basin Transition and Escape Edges 4.2 Monotonic LON (MLON) and Compressed MLON (CMLON) 4.3 Space Metrics 5 Experimental Results 6 Conclusions and Future Work References Dealing with Inconsistencies in ASPIC+ 1 Introduction 2 The ASPIC+ Framework 2.1 Attacks and Defeats 3 Rationality Postulates 3.1 Relation Between Att-conflict-free and Def-conflict-free 4 Related Work and Discussion 5 Conclusion and Future Works References A Grammar-based Genetic Programming Hyper-Heuristic for Corridor Allocation Problem 1 Introduction 2 Problem Description 3 Methods 3.1 Grammar-Based Genetic Programming 3.2 Perturbation Strategies 3.3 Local Search Approaches 4 The Proposed GGPHH 5 Computational Experiments 5.1 Analysis of the Results 5.2 Analysis of the Best Heuristic Created 6 Concluding Remarks and Future Works References Generalising Semantics to Weighted Bipolar Argumentation Frameworks 1 Introduction 2 Preliminaries 2.1 Abstract Argumentation Frameworks (AAFs) 2.2 Bipolar Argumentation Frameworks (BAFs) 3 Semantics for Weighted Bipolar Argumentation Frameworks 4 Results 5 Related Work 6 Conclusion and Future Works References The Use of Multiple Criteria Decision Aiding Methods in Recommender Systems: A Literature Review 1 Introduction 2 Recommender Systems 3 Multiple Criteria Decision Aiding 4 Methodology 4.1 Phase 1: Protocol Definition 4.2 Phase 2: Review Conduction 5 Results 6 Conclusions References Explaining Learning Performance with Local Performance Regions and Maximally Relevant Meta-Rules 1 Introduction 2 Meta-Learning 3 Finding Local Performance Regions in Instance Space 3.1 Definitions 3.2 Finding LPRs 3.3 Finding Maximally Relevant LPRs 3.4 Explaining LPRs Using Meta-Features 4 Experiments 4.1 Finding LHRs: Algorithm 1 4.2 Filtering Meta-Rules: Algorithm 2 4.3 Explaining Single Instances 4.4 Baselines 5 Conclusion References Artificial Intelligence, Algorithmic Transparency and Public Policies: The Case of Facial Recognition Technologies in the Public Transportation System of Large Brazilian Municipalities 1 Introduction 2 The LGPD as an Algorithmic Transparency Instrument 3 Methods 4 Results 5 Discussion 6 Conclusion References Resource Allocation Optimization in Business Processes Supported by Reinforcement Learning and Process Mining 1 Introduction 2 Preliminaries 2.1 Business Process Management and Process Mining 2.2 Markovian Decision Process and Batch Reinforcement Learning 3 Related Work 4 Problem Definition 4.1 Real-World Business Process Context 4.2 Data Preprocessing 4.3 Definition of the Markovian Decision Process 4.4 Application of Reinforcement Learning Algorithms 4.5 Simulation of Policy Application 5 Results and Discussions 6 Conclusions References Exploitability Assessment with Genetically Tuned Interconnected Neural Networks 1 Introduction 2 Related Work 3 Data 4 Enhanced Genetic Algorithm 4.1 Optimizing the Enhanced Genetic Algorithm 5 Neural Network Base Models 5.1 Neural Architecture Search 6 Testing the Enhanced Genetic Algorithm 6.1 On Optimization Benchmark Functions 6.2 On Neural Architecture Search 7 Experiments and Results 8 Conclusion and Further Work References Predicting Compatibility of Cultivars in Grafting Processes Using Kernel Methods and Collaborative Filtering 1 Introduction 2 Recommender Systems in Agriculture 3 Proposed Solution 4 Experiments 4.1 Dataset 4.2 Methods 4.3 Experimental Methodology 5 Results 6 Conclusion References Cross-validation Strategies for Balanced and Imbalanced Datasets 1 Introduction 2 k-fold Cross-validation Partitioning Methods 2.1 Distribution-Balanced Stratified Cross-validation 2.2 Distribution Optimally Balanced Stratified Cross-validation 2.3 Clustering-Based Approaches 3 Experiments 3.1 Estimating the Bias and the Variance 3.2 Defining the Number of Clusters 4 Results and Discussion 4.1 Balanced Datasets 4.2 Imbalanced Datasets 4.3 Running Times 4.4 Cluster-Based Splitters 5 Related Work 6 Conclusion and Future Work References Geographic Context-Based Stacking Learning for Election Prediction from Socio-economic Data 1 Introduction 2 Background and Current Trends 2.1 Problem Definition and Research Challenges 2.2 Related Work 3 Proposed Approach 4 Case Study 4.1 Brazilian Election Data 4.2 Machine Learning Approaches and Algorithms 4.3 Evaluation Measures 4.4 Experimental Setup 4.5 Results and Discussion 5 Conclusion References Author Index