Week 6: Blood Cell Classification

Team Members: Emre Tunç, Muhammed Sezer Berker, Tolga Furkan Güler

Hello everyone, this is our sixth article of series of our Machine Learning Course Project about Blood Cell Classification.To remind, our purpose is classifying the blood cell images and predict the possible disease according to the blood cell we have detected.Some diseases associated with blood cells are as follows: Anemia, Leukopenia, leukocytosis, Platelets.

Types of white blood cell

Last week, we trained our data with VGG-16, AlexNet, ResNet-50 and Cnn model, which we coincided with in a related work.Then we tested the models with test data. We obtained Test-Loss and Test-Accuracy results.We found that the validation and test results overlap and we found that we did not experience overfitting problems.Only in the test results of the AlexNet model, we have observed overfitting so, early stopping process was applied after weights were taken which had the highest validation accuracy in epochs.After examining our results, we decided that VGG-16 is the most successful model for our data set.

This week, we tried a different classification algorithm to improve our model succes rates. We combined our Cnn models with the svm algorithm.Our test results for each Cnn model are as follows.

Confusion Matrix of VGG-16 and CNN Architecture Used in Related Work With SVM

Confusion Matrix of ResNet-50 and AlexNet with SVM

We observed the results and found that the normal cnn algorithm was more successful than using svm.We decided not to use svm because it reduces our success.

This is the last blog of our project.Thank you for reading and for your time.