Machine Learning and Other Soft Computing Techniques: Biomedical and Related Applications

✍ Scribed by Nguyen Hoang Phuong (editor), Nguyen Thi Huyen Chau (editor), Vladik Kreinovich (editor)

• Publisher: Springer
• Year: 2024
• Tongue: English
• Leaves: 254
• Series: Studies in Systems, Decision and Control, 543
• Category: Library

✦ Synopsis

This book contains applications to various health-related problems, from designing and maintaining a proper diet to enhancing hygiene to analysis of mammograms and left-right brain activity to treating diseases such as diabetes and drug addictions. Health issues are very important. So naturally whatever new data processing technique appears, researchers try to apply it to health issues as well. From this viewpoint, Artificial Intelligence (AI) and Computational Intelligence (CI) techniques are no exception: they have been successfully applied to medicine, and more promising applications are on the way. Applications of AI and CI techniques to health issues are the main focus of this book.

Health issues are also very delicate, because human bodies are complex organisms. No matter how interesting and promising are new ideas and new techniques, there is always a possibility of unexpected side effects. Because of this, we cannot apply untested methods to patients, and we first need to test these methods on other less critical applications. Several book chapters describe such applications―whose success paves the way for these methods to be used in biomedical situations. These applications range from human/face detection to predicting student success to predicting election results to explaining the observed intensity of space light.

We hope that this book helps practitioners and researchers to learn more about computational intelligence techniques and their biomedical applications―and to further develop this important research direction.

✦ Table of Contents

Preface
Contents
How to Estimate Unknown Unknowns: From Cosmic Light to Election Polls
1 General Introduction
2 First Case Study: Space Light
3 Second Case Study: Election Polls
4 Possible Explanation
References
Why Bump Reward Function Works Well in Training Insulin Delivery Systems
1 Formulation of the Problem
2 Analysis of the Problem and the Resulting Explanation
References
We Can Always Reduce a Non-linear Dynamical System to Linear—At Least Locally—But Does It Help?
1 Formulation of the Problem
2 Our Answers
References
How to Best Retrain a Neural Network if We Added One More Input Variable
1 Formulation of the Problem
2 Analysis of the Problem
3 Resulting Proposal
4 Experiments
References
Towards a Psychologically Natural Relation Between Colors and Fuzzy Degrees
1 Formulation of the Problem
2 Towards the Desired Natural Relation
3 Discussion
References
Algebraic Product Is the only ``And-Like''-Operation for Which Normalized Intersection Is Associative: A Proof
1 Formulation of the Problem
2 Main Result
References
High Potential Negative Sampling for Drug Disease Association Prediction
1 Introduction
2 Related Work
3 The Method
4 Experiments
5 Conclusions
References
Cognitive States Prediction with KNN and TomekLinks
1 Introduction
2 Related Work
3 The Method
4 Experiments
5 Conclusions
6 Appendix
References
Health Digital Twins with Clinical Decision Support and Medical Imaging
1 Introduction
2 Methods
3 Results
4 Discussion
5 Conclusion
References
Promoting STEM-Integrated Learning Through Engineering Design: High School Students' Automatic Hand Washers
1 Introduction
2 Related Work
3 Methods
3.1 Description of Learning Tasks
3.2 Building Rubric to Evaluate STEM Activities
4 Experiment
4.1 Experimental Object and Process
4.2 Evaluation
4.3 Discussion
5 Conclusion
References
KNN-SMOTE: An Innovative Resampling Technique Enhancing the Efficacy of Imbalanced Biomedical Classification
1 Introduction
2 Related Work
3 The Method
4 Experiments
4.1 Datasets
4.2 Evaluation Measures
4.3 Classification Imbalance Learning Results
5 Conclusions
References
Human Detection in Video for Security Surveillance Systems
1 Introduction
2 Related Work
3 Proposed Approach
3.1 Yolov7 Detector
3.2 Sequential Model
3.3 VGG16 Transfer Learning
4 Experiments
4.1 Data Set
4.2 Experimental Results
5 Conclusion
References
Fake Face Detection with Separable Convolutions
1 Introduction
2 Related Work
3 Methods
3.1 Dataset
3.2 Deep Learning Architectures
3.3 Separable Convolutions
4 Experimental Results
4.1 Environment Setting
4.2 Results
5 Conclusion
References
A Classification System of Mammograms Based on Convolutional Neural Networks
1 Introduction
2 System Design
3 Data Collection and Labeling
4 Data Pre-processing
5 Model Training and Evaluation
6 Conclusions
References
OAGRE: Outlier Attenuated Gradient Boosted Regression
1 Introduction
2 Method
2.1 Implementation
2.2 Evaluation
3 Results
4 Conclusion
References
Improve the Effectiveness of Predicting Student Dropouts Based on Deep Learning and SMOTE Models
1 Introduction
2 Related Work
3 The Method
3.1 Datasets
3.2 Data Imbalance Preprocessing
4 Experiments
5 Conclusions
References
Data Processing and Feature Engineering for Stock Price Trend Prediction
1 Introduction
2 Data Collection
3 Data Preparation
4 Feature Engineering
5 Model Development
6 Experimental Results
6.1 Feature Engineering and Non-feature Engineering
6.2 Predicting Future Data
6.3 Comparison with Results from Related Works
7 Summary
References
Distributed Computing in Training Machine Learning Models
1 Introduction
2 Distributed Computing Overview
2.1 Data Parallelism Versus Model Parallelism
2.2 Decentralized Asynchronous Systems
3 Proposed Distributed Computing Method
3.1 Communication Process with Socket Library in Python
3.2 Data Parallelism Model Design and Deployment
4 Experimentations
4.1 Experimentation Results
4.2 Insights and Experiences
5 Discussion on Future Works
6 Conclusion
References
Fruit Calorie Determination System for Dieters and Athletes Using Deep Learning
1 Introduction
2 Related Work
3 Proposed Approach
4 Experiments
4.1 Dataset
4.2 Experimental Results
5 Conclusion and Future Works
References
An Approach to Instrumental Song Classification Utilizing Spectrogram and Convolutional Neural Networks
1 Introduction
2 Related Work
3 Methods
3.1 Data Collection and Division of Songs
3.2 Transforming Audio Signal to Image with Spectrogram
3.3 The Networks for Song Recognition
4 Experimental Results
4.1 Experimental Setup
4.2 The Length of Pieces Extracted from the Song Can Affect the Song Detection Performance
4.3 Data Augmentation on Songs
4.4 Classification Algorithms Comparison
5 Conclusion
References
Heterogeneous Transfer Learning Using Pre-trained Feature Mapping and Exchange
1 Introduction
2 Related Works
3 Proposed Method
3.1 Stage 1: Matching
3.2 Stage 2: Convolutional Transfer
3.3 Stage 3: Fully-Connected Transfer
3.4 Training with Feature Exchange
4 Experimental Results
4.1 Setup
4.2 Using Cifar10, Cifar100, and PetImages Datasets
4.3 Ablation Study
5 Conclusion
References
Usually, Either Left and Right Brains Are Equally Active or Only One of Them Is Active: First-Principles Explanation
1 Formulation of the Problem
2 Definitions and the Main Result
3 Proofs
References