Machine Learning techniques to tackle the problem of “Detection of Pneumonia in Medical Images and Game Learning (6 pages)

Rules:

1. A well-chosen set of classifiers and reinforcement learning agents is provided, which can be applied to the test data.

2. Even when the results might not (very) high, they show ambition or promise in the proposed solutions.

3. A well written, well organised and crafted report is provided. The report presents a very logical and convincing discussion of the selection of machine learning tools to create the trained models

4. There is convincing justification regarding the decisions and choices made, which are supported by relevant references from academic literature.

Ex of algorithms:

CNN models (Deep Learning): Transfer learning GoogLeNet, ResNet-18, and DenseNet-121

ML Algorithms: classification models, such as decision trees, support vector machines, and logistic regression

Ex Code

https://github.com/Rohit-Kundu/Ensemble-Pneumonia-Detection/blob/main/probability_extraction.py

Evaluation metrics to find out the accuracy

Confusion metrics (Evaluation metrics):

four standard evaluation metrics are: accuracy (Acc), precision (Pre), recall (Rec), and f1-score (F1)

[ “True Positive,” “False Positive,” “True Negative,” and “False Negative.”]

Ex Code

https://github.com/Rohit-Kundu/Ensemble-Pneumonia-Detection/blob/main/ensemble.py

Ex’s of Evaluation metrics & ROC AUC Curve

Ex of Accuracy metrics:

Accuracy score Comparision with different algorithms: