1) Utilizing Machine Learning in Healthcare: Advancements and Applications: 

​

• Artificial Intelligence in Chronic Kidney Disease: Uncovering Patient Phenotypes (Generative Adversarial Networks)

​

     Description: ​Chronic kidney disease (CKD) is a major public health problem that is associated with a high risk of adverse clinical events and high healthcare costs. CKD is a progressive condition that gradually worsens over time. It affects more than 10% of the world's population, amounting to more than 800 million people (P. Kovesdy, 2022). In 2019, treating Medicare beneficiaries with CKD cost 87.2 billion dollars in the US (cdc.gov). In this project, we aim to propose a novel approach utilizing GANs to identify the disease during its initial stages. Early diagnosis is crucial, as it can help to prevent or slow the progression of the disease.

​

 

• Firefighter Injury and Entrapment in Urban Firefighting Operations (KNN, SVM, NN, DT, RF)

     

     Description: ​Urban fires pose a significant threat in terms of property damage and potential loss of life. Firefighters play an important role in managing these incidents, so their safety is a top priority for fire departments and emergency responders. In this project, we have modified two pre-existing stages of fire time by adding new features. The main objective of the modifications was to enable early prediction for the occurrence of injury and for entrapment of firefighters during urban firefighting operations.

​

​

2) Advancing Manufacturing Industry through Machine Learning Applications:

​

• A Sequential and Active Learning Framework to Optimize Multi-Objective Manufacturing Decisions

​

Description: Optimizing the manufacturing of advanced materials and products requires identifying the ideal recipe or processing conditions. Conducting experiments to generate a Pareto front can be expensive and time-consuming. To gain comprehensive insights while minimizing costs, strategically determining the optimal data collection location is crucial. This project presents a novel data-driven Bayesian optimization framework that utilizes sequential learning to efficiently optimize complex systems with multiple conflicting objectives.

​

​

• An Improved CTGAN-based Surprise-guided Sequential Learning Framework in Predicting the Melt Pool Geometry

​

Description: Metal Additive Manufacturing (MAM) revolutionizes manufacturing with complex geometries and reduced waste. However, industry adoption is slow due to quality concerns. We propose a novel sequential learning approach to predict melt pool geometry, comparing it with EI-based Bayesian Optimization and six ML techniques. Integrating a CTGAN model enhances performance in resource-constrained environments, aiming to ensure high-quality parts.

​

​

3) Binary Metaheuristic algorithms for feature selection / Feature Selection algorithms in

 

High-Dimensional Data 

​

Description: High-dimensional datasets usually contain extra and unrelated features that increase complexity and reduce matching learning algorithm accuracy [Alweshah et al.]. Feature selection in datasets can considerably improve the performance of matching learning algorithms by reducing the learning model creation time and increasing the accuracy of the learning process [Sihwail et al.] Recently, meta-heuristic and evolutionary computation algorithms have been used as envelopment-based methods for exploring vast search spaces [Abd Elaziz et al.].Binary Genetic algorithm and Binary PSO algorithm are used here.

​

4) Cutting-Edge AI and Optimization Algorithm Development

​

• Presenting An Improved Feature Selection Algorithm for High-Dimensional Data

• Presenting An Improved Algorithm for Global Optimization

​

Description: Introducing an improved feature selection model based on the Group Teaching Optimization Algorithm, and a novel and improved version of the Farmland Fertility algorithm for global optimization.