alecruces / OptiRice

Exploring optimization methods like Gradient Descent and BCGD in semi-supervised learning for rice seeds classification

Gradient Descent and BCGD Methods with an application on Rice Seeds classification

Description

Semi-supervised learning is a technique that deals with data that has both labeled and unlabeled instances. The goal is to use the information from the labeled data to predict the labeling of the unlabeled data. In this report, three different optimization Methods for semi-supervised learning on a synthetic dataset and a real dataset about rice seeds are used. In both datasets, just a small fraction of data with labels is kept. The performance of the Methods is evaluated based on their accuracy and time.

Keywords

Optimization Methods, Semi-Supervised Learning, Rice Seeds Classification

Data Rice Type Data Set

Rice type classification (kaggle.com)

Methods

• Gradient Descent
• BCGD randomized rule
• BCGD Gauss-Southwell rule

Software Python

• Numpy
• Pandas
• Matplotlib

Files

• Code: code.ipynb
• Report: Report.pdf