Performance Evaluation of non linear Estimators using Genetic Programming

Our project is an R&D project we have to do research first then to get optimized results and after that, we will design hardware for it. The summary of the project named "performance evaluation of non-linear estimators using genetic programming" is, therefore, to improve the p

Project Title

Performance Evaluation of non linear Estimators using Genetic Programming

Project Area of Specialization

Artificial Intelligence

Project Summary

Our project is an R&D project we have to do research first then to get optimized results and after that, we will design hardware for it. The summary of the project named "performance evaluation of non-linear estimators using genetic programming" is, therefore, to improve the prediction accuracy using non-linear models. Then evolving that model using genetic programming and comparing the results, the model having better prediction will be then lead to further hardware work. In particular, it is a biomedical engineering project used to detect breast cancer(mass to be either malignant or benign). Breast Cancer is the second most dangerous disease who has taken the lives of many innocent people after lung cancer. Among women, it is the most widely spread disease according to a recent survey its rate is increasing 0.4% per year which is a severe threat for seven and half billion people. Once we get the generic form and better results we will design hardware for it that will help doctors in their diagnosis.

Project Objectives

The sole objectives of the project can be better understood if we number it

1)To select the model from so many non-linear models(we have chosen ANN) and optimize it.

2)Then using genetic programming to evolve that non-linear model and see its behavior.

3)Then cross-validation of the upper two.

4)Designing hardware using the research work obtained.

These 1 to 4 steps are due to helping those patients in the early detection of their disease and to treat it well in its early stages and to save their lives.

Project Implementation Method

We have chosen Wisconsin Breast Cancer Diagnosis Dataset to be our input dataset. Then we will split this dataset to testing and training and will train RNN recurrent neural network over it and then will test the accuracy for the testing data sets. Then we will use the same Fine needle aspiration data that is Wisconsin breast cancer data sets in the split form to evolve it for the neural network. Once we get the best generic form we will design hardware for it to take it to reality I.A.

Benefits of the Project

United Nations sustainable development goal number 3 is about good health. It is targeting that to make good health common and to reduce the number of cancer patients in the future. Moreover, it is to help doctors by diagnosing patients up to more accuracy than the traditional way of diagnosing cancer which is only 79 % for an expert doctor as there are so many features to take into consideration and chances are there to miss some of them or negate it. While using machine learning algorithms the accuracy can be enhanced to a satisfactory level(more than 95%), which is better than 79%.

And having diagnosed righty in its early stage will have time to cure it and to avoid its fatality.

We are also aiming to make it available to distant areas using the Internet of things devices which will be connected with the main hospitals and the diagnosis will be done in real time by the specialist doctors.

Technical Details of Final Deliverable

The problem with genetic programming is that it takes to much time as we haven't any GPU that is used to do parallel processing thus decrease the time if a single fold takes 5 minutes then if we have used five folds cross-validation and we have at least 1000 generation then it will take about 5*1000*5=416 hours to complete just 1000 generation and hence we can't achieve the best-optimized result that will reduce the fatality. 

For Hardware design we have to take photos using Microwave sensor camera and using those result we will program to act upon those pictures.

We will make it able to connect it to the main server that will help the distinct area of our country to reach specialized  doctors in a convenient way rather than to travel hundreds of kilometers and the result will also be stored in a database to make it present to all over the world to act upon it and to increase the performance I.A.

Final Deliverable of the Project

HW/SW integrated system

Type of Industry

Medical , Health

Technologies

Artificial Intelligence(AI), Internet of Things (IoT)

Sustainable Development Goals

Good Health and Well-Being for People

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com