0% completed

All LessonsFree Lessons (4)

Introduction to the Course

Course Overview Who Can Take This Course?

Getting Started

Overview of AI, ML, DL, and Supervised/Unsupervised Learning Image Classification Image Classification Using Python Programming Data Types The Model-Centric Approach vs. the Data-Centric Approach Quiz: Artificial Intelligence Basics

Understanding Noisy Data, Label Noise, and Its Types

Noisy Data and Label Noise Simulating Unbiased Mislabeling Using Python Programming Simulating Biased Mislabeling Using Python Programming Quiz: The Fundamentals of Noisy Data, Label Noise, and Its Types

Introduction to Convolutional Neural Network (CNN)

Image Classification Using Convolutional Neural Networks Input Layer and Convolution Layer Implementation of Pooling Layers Using Python Programming Hyperparameter Tuning Quiz: Fundamental Concepts of CNN Architecture

Project

Cats vs Dogs Classification with Convolutional Neural Networks

Performance Comparison of Mislabeled and Clean Dataset

Implementing CNNs for Image Classification in Python Unbiased Mislabeling in Image Classification Using CNNs Biased Mislabeling in Image Classification Using CNNs Dealing with Mislabeled Datasets Using Pretrained Models Quiz: Evaluating the Performance of Mislabeled and Clean Datasets

Dealing with Imbalance Dataset

Imbalanced Datasets Methods for Transforming Imbalanced Data into Balanced Data Dealing with Imbalanced Datasets in Python Programming Quiz: Essential Concepts for Dealing with Imbalanced Datasets

Mini Project

Gauge the Impact of Imbalanced and Mislabeled Datasets

Course Assessment

Comprehensive Quiz

Wrap Up

Appendix

Project

Dealing With Small Datasets In ML

Deal with Mislabeled and Imbalanced Machine Learning Datasets/

Gauge the Impact of Imbalanced and Mislabeled Datasets

Gauge the Impact of Imbalanced and Mislabeled Datasets

This project assesses the knowledge acquired in this course. The selected dataset for this project is the MNIST handwritten digit dataset, which is renowned for its simplicity and is commonly used in ML. This dataset comprises 70,000 images, each containing handwritten digits ranging from 0 to 9. These images are categorized into two sets: 60,000 for training and 10,000 for testing. Each image is composed of 28 × 28 grayscale pixels.

MNIST handwritten digit dataset

MNIST handwritten digit dataset

This project contains two comprehensive sections that cover the entire course. The first section gauges the impact of mislabeled datasets on the performance of ML models. The second section evaluates the effect of imbalanced datasets on the performance of ML models and explores strategies for effectively addressing imbalanced data. Together, these sections will help us understand data-related challenges and essential solutions for handling these challenges in ML.