Special Issue on “Machine Learning & Mathematical Modeling: A New Approach”
Aims and Scope
The primary scientific pattern for modeling real-world processes and natural phenomena includes removing experience knowledge, formalizing this information, and empirically testing the model. Physical rules, chemical processes or complex behaviors are represented e.g. using differential equations, whereas verification can appear as forecasting future procedural circumstances. With the increasing accessibility of massive perception datasets, this real-world modeling trend is being challenged by a statistical Machine Learning (ML) scheme that straightforwardly shapes information to construe forecast, bypassing the human formalization of process details. This is the hotspot problem we address: how might general knowledge obtained from phenomenon modeling trend help in constructing efficient ML algorithms? Both ML and mathematical models give specific advantages in this case, and one might combine the two in an ideal environment. Such a solution will have the same design and low mathematical model computational capacity, but might add some robustness by machine learning. This could take the form of an additional cushion depending on different causes, such as the amount of anticipated disruptions, or the steady decrease in a players success over their career or a long season.
This is precisely how logistics utilizes ML and statistical simulations. ML and statistical simulation are also in the scheduling world. Model-based ML is all the theories regarding a physical or natural systems mathematical models that are constructed specifically as a type of differential or algebraic equation. A model essentially consists of this set of theories, stated in a precise mathematical schema. These theories include the number and types of parameters in the model domain, which parameters impact each other and the utility of varying one parameter on another.
For eg, we make a model that attempts to solve an uncomplicated murder mystery. The issue theories include the file of offenders, the possible murder arms, and the tendency of multiple suspects to prefer particular weapons. This dilemma is then used to construct a model-explicit algorithm to solve the explicit ML scheme. Model-based ML can be used to solve almost every natural or real-world phenomenon, and its commonly used approach ensures that you do not need to learn a large range of ML algorithms and methods.
This special issue aims to offer researchers the opportunity to deepen the linkages between mathematical modelling of natural phenomena and ML algorithms and to shed light on this current hotspot.
Topics of Interest: Topics to be discussed in this special issue include (but are not limited to) the following:
v Machine learning hybrid approach for accurate mathematical modeling
v Artificial Neural Networks for Probabilistic models
v Temporal models with recurrent neural networks
v Fractional-order gradient descent approach to signal processing and machine learning
v Deep neural networks for unstructured problems with massive data
v Fractional-order steepest gradient descent
v Evolutionary process of fractional back-propagation neural networks
v Combining physically-based modeling and deep learning
v Incorporating machine learning and multiscale physical modeling
v Artificial life simulations
v Integrating of economic models and ML algorithms to predict economic trends
v Identifying system dynamics by machine learning
v K-mean clustering in combination with epidemic diseases modeling to simulate the dynamical behaviours
Dr. Sathishkumar Karupusamy, Bharathiar University, Tamilnadu, India. Email: firstname.lastname@example.org
Dr. Kandappan Balasubramanian, School of Hospitality, Tourism and Events - Taylor’s University, Malaysia . Email: Kandappan.email@example.com
Dr. Chockalingam Aravind Vaithilingam, Taylors University, Malaysia. Chockalingamaravind.firstname.lastname@example.org / email@example.com