/Machine_Learning_idu

Two case studies: effects of changing the learning rate on model perfomance for image classificaiton, and cardiac failure prediction using clinical data

Primary LanguageJupyter Notebook

1. Changing the learning for better performance--case study

This repository is a final Project: The project discusses the effects of changing the learning rate on image classification modelling performance. Two algorithms were compared through two different codes, one with fixed learning rate and one with changing learning rate. The comparison was made in terms of accuracy performacne and loss performacne for a multiclass image classification task. The popular CIFAR100 dataset was used for validation of the results.

  • The codes can be found in "CIFAR-100" Folder
  • Explanatory video can be found here.

2. Prediction of Cardiac Failure

DOI:10.1109/SACI.2018.8440931

  • The dataset used was obtained from a hospital in Hungary. The platform used for the work is Rstudio.
  • The work is in folder "Hospital-Data-Project"

Citation

If you use this work, consider citing the paper as:

@INPROCEEDINGS{8440931,
author={Morani, Kinan and Eigner, György and Ferenci, Tamás and Kovács, Levente and Naci Engin, Seref},
booktitle={2018 IEEE 12th International Symposium on Applied Computational Intelligence and Informatics (SACI)},
title={Prediction of the Survival of Patients with Cardiac Failure by Using Soft Computing Techniques},
year={2018},
volume={},
number={},
pages={000201-000206},
doi={10.1109/SACI.2018.8440931}}