Bachelor’s thesis

Thesis can be found here: bs_thesis_en.pdf
Диплом расположен тут: bs_thesis_ru.pdf

This repository contains the code used to accomplish N.S. Detkov’s bachelor’s thesis.

Topic: Influence of the label smoothing of pseudo-labeled data on training Convolutional Neural Networks

Steps to reconstruct the solution:

1. Download the data from the site and extract to the input/ folder
2. (Optional) Exploratory Data Analysis

   cd src
   juputer lab

   Further, you need to open the EDA.ipynb file — it contains an overview and visualization of the presented data
3. Data pre-processing, which generates reduced resolution images and splits the training data into folds

   cd src
   bash preprocess.sh

4. Launch of the training process

   python train_infer.py -c exp_train_02.yaml

5. Review the statistics on the predictions of the test set

   python show_prediction_stats.py -f exp_train_02.csv

6. Generation of pseudo-labeled datasets for the fine-tuning

   ipython Create_Datasets.ipynb

7. Generation of pseudo-labeled datasets applying label smoothing, each of which is "experiment"

   ipython Create_Experiments.ipynb

8. Launch fine-tuning experiments on pseudo-label with label smoothing

   ipython Run_Experiments.ipynb

9. Overview of the experiments results

   juputer lab

   Further, you need to open the Analyse_Experiments_Results.ipynb file — it contains an overview and visualization of the presented data