In this research, I have shown the root causes of Machine Learning fairness issues are the simplistic assumptions made to enable supervised learning.
Machine Learning Fairness: Supervised or Unsupervised Learning?
Article Name: Supervised or Unsupervised Learning? Investigating the role of pattern recognition assumptions in the success of binary predictive prescriptions
Published in Neurocomputing 28 April 2021
Decision-optimizing Supervised Learning
Optimum profit-driven churn decision making: innovative artificial neural networks in telecom industry
This article showcases how Supervised Learning at the level of model construction can be adjusted so the decisions that are made based on the models are optimizing profit.
Novel Machine Learning Algorithm
Self-organizing and error-driven (SOED) artificial neural network for smarter classifications
I have developed and published this new ML algorithm that effectively combines both supervised and unsupervised learning algorithms of Artificial Neural Networks. By effectiveness, in this article, I refer to the quality of decision-making and not speed or accuracy.
Novel Metaheuristics
Fluid Genetic Algorithm (FGA)
Genetic Algorithm (GA) (GA) suffers from premature conversion and getting stuck at the local optima. This effort improved GA by making the chromosomes more fuzzy-like and probabilistic. The advantage of this approach is its significantly less premature conversion but the disadvantage is that the chromosome design is not as flexible as the general GA.
