Gain insights into dealing with mislabeled and imbalanced machine learning datasets. Learn to analyze effects, measure and recover from noise, and interpret results to avoid bias.

Impact of Mislabeling in Machine Learning.png

jupyter.tar.gz

Machine learning models depend thoroughly on the dataset quality they are trained on. The model’s performance deteriorates significantly due to noisy datasets. One primary source of noise is mislabeling. Labeling is a costly, time-consuming, and error-prone stage in the machine learning pipeline. Data, if not correctly labeled, can introduce bias and inaccuracies into machine learning models.

This course offers hands-on experience in analyzing the effects of mislabeled datasets on machine learning models, especially convolutional neural networks. It emphasizes the modern data-centric perspective in machine learning. Eventually, it teaches how to measure and recover from noisy datasets.

After completing this course, you will be skilled at handling imbalanced datasets and be able to interpret results fairly to avoid bias toward minority classes. Having such skills is vital in machine learning and important for both industry and academia.

image

Deal with Mislabeled and Imbalanced Machine Learning Datasets

# Import necessary libraries
import keras
import numpy as np
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D, Flatten, Dense
import matplotlib.pyplot as plt
from tensorflow.keras.optimizers import Adam

# Load the MNIST dataset
(x_train, y_train), (x_test, y_test) = mnist.load_data()

# Define the percentages for training and testing data
train_percentage = 0.25  # 15,000 images for training
test_percentage = 0.2   # 2,000 images for testing

# Calculate the number of samples based on percentages
total_train_samples = len(x_train) 
total_test_samples = len(x_test)
train_samples = int(train_percentage * total_train_samples)
test_samples = int(test_percentage * total_test_samples)

# Distribute the data based on percentages
x_train = x_train[:train_samples]
y_train = y_train[:train_samples]
x_test = x_test[:test_samples]
y_test = y_test[:test_samples]

# Define the percentage of mislabeled images
mislabel_percentage = 10

# Compute the number of images to mislabel
num_mislabeled = int(len(y_train) * mislabel_percentage / 100)

# Randomly select images to mislabel
index = np.random.choice(len(y_train), size=num_mislabeled, replace=False)

# Generate new labels for the mislabeled images
new_labels = np.random.randint(0, 9, size=num_mislabeled)

# Create a copy of the original training set and replace the selected images with the mislabeled ones
x_train_mislabeled = np.copy(x_train)
y_train_mislabeled = np.copy(y_train)
x_train_mislabeled[index] = x_train[index]
y_train_mislabeled[index] = new_labels

Explore how an unbiased mislabeled dataset affects the performance of a CNN model.

Introduction to the Course

Getting Started

Understanding Noisy Data, Label Noise, and Its Types

Introduction to Convolutional Neural Network (CNN)

Cats vs Dogs Classification with Convolutional Neural Networks

Performance Comparison of Mislabeled and Clean Dataset

Dealing with Imbalance Dataset

Gauge the Impact of Imbalanced and Mislabeled Datasets

Comprehensive Quiz

Wrap Up

Appendix

Dealing With Small Datasets In ML

Unbiased Mislabeling in Image Classification Using CNNs

Implementing unbiased mislabeling

Step 1: Importing libraries

Step 2: Loading and creating an unbiased mislabeled dataset