Collaborate
Publications Require Collaborators
A Data-Theoretic Approach to Identifying Violent Facial Expressions in Social Crime Contexts
The paper proposes an innovative data-theoretic approach to identify violent facial expressions in the context of social crimes. Recognizing the crucial role facial expressions play in understanding human actions and intentions, the authors design an automated system using Convolutional Neural Networks (CNN). This system aims to detect whether an individual has any intention to commit a crime by analyzing facial expressions. Notably, the study emphasizes the influence of geographic regions on patterns of violent behavior. The authors introduce a novel method that efficiently identifies crime intentions or violent behavior using minimal facial expression data. Unlike traditional methods using image features, the proposed model employs an automated feature selector CNN, capturing precise facial expressions for training and achieving more accurate predictions. The focus is on a specific geographic region to represent the distinct facial patterns associated with violence and crime.
Objectives and Novelty
Methodological Developments, Verifications, Challenges in Hyperparameters Tuning, Convergences, Prediction on New Data. Training on minimum data.
Methods and Procedure
Data Collection
Collect facial images representing individuals before committing crimes, categorized into suspects.
Gather general facial photos from various sources, including non-suspected individuals engaged in normal activities.
Select faces of actors portraying crime scenes for experimentation.
Model Architecture
Implement a Convolutional Neural Network for prediction purposes.
Experiment with different model configurations, including optimizers, loss functions, and activation functions.
Evaluate performance in terms of accuracy and loss, considering the small dataset and noisy images.
Key findings
The study demonstrates the potential of CNNs in accurately identifying violent facial expressions.
Geographic regions significantly influence patterns of violent behavior and crime intentions.
The proposed method efficiently identifies crime intentions using minimal facial expression data.
An automated feature selector CNN enhances the accuracy of capturing and predicting target facial expressions.
Model performance varies with different configurations, requiring careful selection based on the dataset characteristics.
Collaboration Demands any contribution in publishing the research work either by contributing the APC, or by contributing to the development of the work.
Modeling the Spread of COVID-19 Among Doctors from the Asymptomatic Individuals
The focus is on developing a non-linear system of five differential equations based on the tendency of doctors to be infected. The objective is to analyze the transmission of COVID-19 from asymptomatic populations to doctors. The model is evaluated, considering parameters like the basic reproduction number, equilibrium points, and stability analysis. The research aims to understand how the disease spreads, affecting doctors and impacting the overall medical service system.
Objectives
Formulate a non-linear system of differential equations for COVID-19 transmission involving asymptomatic individuals and doctors.
Investigate the basic reproduction number, equilibrium points, and stability analysis of the model.
Analyze the impact of disease transmission rates on asymptomatic and infected populations.
Understand the dynamics of the medical service system as doctors get infected and the disease spreads.
Methods and Procedure
Develop a five-compartmental model representing susceptible, asymptomatic, infected, doctors, and recovered populations.
Formulate a set of non-linear ordinary differential equations (ODE) based on the interactions between different compartments.
Conduct a positivity analysis to validate and establish well-posedness of the model.
Calculate equilibria, basic reproduction ratio, and perform stability analysis at disease-free and endemic equilibrium points.
Utilize numerical simulations using the Runge–Kutta-Fehlberg method and MATLAB programming language.
Explore the impact of different parameters, such as disease transmission rates and recovery rates, on the model through graphical illustrations.
Key Findings
The model reveals that the asymptomatic population and the number of infected individuals increase with higher disease transmission rates.
Doctors are more prone to infection as the disease transmission rate rises, leading to a decline in the overall medical service system over time.
Numerical simulations illustrate the dynamics of susceptible, asymptomatic, infected, doctor, and recovered populations over a 15-week period.
The basic reproductive ratio helps predict whether the disease will persist or die out, indicating the severity of the COVID-19 spread.
Research Gaps
The study lacks discussion on specific control policies or interventions to mitigate the impact on doctors and the medical service system.
There is no exploration of external factors, such as vaccination or public health measures, that could influence the model dynamics.
The research does not address the potential variations in disease transmission rates based on different regions or healthcare settings.
Future Developments
Future research could focus on incorporating control strategies into the model, such as vaccination campaigns or social distancing measures, to assess their effectiveness.
Investigate the impact of external factors and regional variations on the transmission dynamics of COVID-19 among doctors.
Extend the model to consider the interaction between multiple populations and explore the interconnectedness of disease spread in various settings.
Publisher: Springers Denmark, Book Chapter.
This Research is included in the WHO COVID-19 Research Database, Anyone interested to extend this study to the indicated directions or any other proposed directions will get guidelines and overall support.
Modeling and optimal control applied to reduce the effects of greenhouse gases emitted from the coal-based power plant in Bangladesh
Objectives
Investigate the impact of coal-fired power plant emissions, particularly CO2, on the Sundarbans.
Develop a compartmental mathematical model to analyze greenhouse gas dynamics and propose optimal control strategies.
Apply Pontryagin's maximum principle to formulate an optimal control problem.
Assess the effectiveness of control strategies, considering factors like system damage and absorbers.
Compare and evaluate the cost-effectiveness of different control systems.
Provide recommendations for minimizing air pollution from coal-based power plants.
Methods and Procedure
Mathematical Model Formulation:
Develop a model based on previous work, introducing two control variables.
Propose a chemical reactor system in the power plant and a tree farming project for afforestation.
Consider factors like pollutants' absorption by trees, transportation rates, and system damage.
Optimal Control Strategy:
Apply Pontryagin's maximum principle to formulate an optimal control problem.
Introduce control variables related to chemical reactions and afforestation.
Consider the impact of pollutants on the Sundarbans and Khulna-Bagerhat areas.
Numerical Simulations:
Simulate the model with five different control strategies.
Evaluate the performance index scores and objective function values for each strategy.
Compare control systems by assessing their costs and effectiveness in reducing pollution.
Key Findings
A New Mathematical Model to Study the GHG emissions and concentration dynamics near mangroves.
Combination of Numerical Simulation and Optimal Control Theory allows others to use this model and experiment different control measures according their demands.
Future Developments
Conduct further research on the specific chemicals and mechanisms involved in the proposed chemical reactor system.
Explore the sensitivity of the model to parameter variations and uncertainties for robust control strategies.
Investigate the economic feasibility and practical implementation of the recommended control measures.
Extend the study to consider the long-term effects of control strategies on the Sundarbans and surrounding areas.
Publisher: Heliyon, Cell Press (Q1) Section: Energy. Mentioned Developments will be supported.