A Python Package for Deep Imbalanced Learning

the DL methods used are : DNN, CNN, LSTM, GRU, CNN-LSTM, CNN-GRU, CNN-BiLSTM, CNN-BiGRU, the dataset used is : imbalanced NSL-KDD (KDDTrain+, KDDTest20+)

Alzheimer’s Disease (AD) is a neurological brain disorder marked by dementia and neurological dysfunction that affects memory, behavioral patterns, and reasoning. Alzheimer’s disease is an incurable disease that primarily affects people over the age of 40. Presently, Alzheimer’s disease is diagnosed through a manual evaluation of a patient’s MRI scan and neuro-psychological examinations. Deep Learning (DL), a type of Artificial Intelligence (AI), has pioneered new approaches to automate medical image diagnosis. The goal of this study is to create a reliable and efficient approach for classifying AD using MRI by applying the deep Convolutional Neural Network (CNN). In this paper, we propose a new CNN architecture for detecting AD with relatively few parameters and the proposed solution is ideal for training a smaller dataset. This proposed model successfully distinguishes the early stages of Alzheimer’s disease and shows class activation maps as a heat map on the brain. The proposed Alzheimer’s Disease Detection Network (ADD-Net) is built from scratch to precisely classify the stages of AD by decreasing parameters and calculation costs. The Kaggle MRI image dataset has a significant class imbalance problem and we exploited a synthetic oversampling technique to evenly distribute the image among the classes to prevent the problem of class imbalance. The proposed ADD-Net is extensively evaluated against DenseNet169, VGG19, and InceptionResNet V2 using precision, recall, F1-score, Area Under the Curve (AUC), and loss. The ADD-Net achieved the following values for evaluation metrics: 98.63%, 99.76%, 98.61%, 98.63%, 98.58%, and 0.0549 for accuracy, AUC , F1-score, precision, recall, and loss, respectively. From the simulation results, it is noted that the proposed ADD-Net outperforms other state-of-the-art models in all the evaluation metrics.

"Google creates advertising algorithms, not information algorithms."― Safiya Umoja Noble

The code proposes various novel loss functions to train the DL models and construct their ensembles to improve performance in a class-imbalanced multiclass classification task using chest radiographs

Addressing challenges like imbalanced data and evolving fraud techniques. Employed various ML algorithms (Decision Tree, Random Forest, etc.) & DL techniques (CNN) for improved accuracy. Demonstrated CNN's effectiveness with 20 layers achieving 99.956% accuracy. Dataset: https://www.kaggle.com/datasets/mlg-ulb/creditcardfraud

🌿 Deep Learning project classifying rare species from images using the BioCLIP dataset. Features transfer learning (ConvNeXtBase), innovative data cleaning with CLIP (zero-shot), and imbalance handling. Developed for the DL course at NOVA IMS.

Sentiment classification of an imbalanced data set with text data using sklearn and keras (ML and DL)

ml,dl on dataset, methods to improve the perforamnce on imbalanced dataset

Brain Stroke Prediction- Project on predicting brain stroke on an imbalanced dataset with various ML Algorithms and DL to find the optimal model and use for medical applications.

A_DL_Model_for_Network_Intrusion_Detection_with_Imbalanced_Data

A simple Python module for conducting imbalance mitigation experiments in DL

An in-depth analysis and summary of machine learning techniques for solar cell defect detection using electroluminescence images. This project explores the effectiveness of various ML and DL models in classifying solar panel defects, addressing challenges of imbalanced data, and providing practical insights for industrial applications.

This project uses Generative Adversarial Networks (GANs) to enhance diabetes prediction accuracy by generating synthetic patient data to address class imbalance. Multiple ML and DL algorithms—including Graph Neural Networks—were trained on the augmented dataset, achieving up to 97% accuracy, with a 40% improvement in model performance. Built using

This project develops a DL model for skin lesion classification using EfficientNetV2S enhanced with ECA and SE attention mechanisms. The model is trained on the HAM10000 dataset to detect three skin cancer types: BCC, MEL, and NV and uses Focal Loss to handle dataset imbalance. It achieves up to 97% training accuracy and 93–97% validation accuracغ

The price of a car depends on a lot of factors like the goodwill of the brand of the car, features of the car, horsepower and the mileage it gives and many more. Car price prediction is one of the major research areas in machine learning. In this project, I am going to Apply Data science techniques to Undertand the core features of the data and build a Machine Learning and deep learning models to predict the car sale price. Projrct content and sections - Business problem Identification - Explorative Data analytics - Data preparation & Cleaning - Data Visualization - Data Understanding and Interprining - Feature selection & Feature Enginering - Understand the Features - Select core Features - Handling Imbalance Data - Encode the catagorical data - Sacle down the data (Apply one hot encoding) - Build MAchine learning models - Split The data - Build the model - Evaluvate the ML model accuracy (1st round) - hyperparameter tuning - Evaluate the ML model accuracy (2st round) - Build Deep Learning Model - Evaluate the DL model accuracy

Deep learning for collateral evaluation in ischemic stroke with imbalanced data

Imbalanced dataset handling in Deep Learning

"Google creates advertising algorithms, not information algorithms."― Safiya Umoja Noble

simple example of handling class imbalance in DL techniques

Handling imbalanced Bank customer churn dataset in Deep Learning

Real-time UPI fraud detection using ML and DL (Logistic Regression, SVM, Random Forest, CNN, LSTM) with high accuracy on imbalanced data.

Crash course on Introduction to AI and its medical applications. Tutorials with focus on intro to DL, overfitting, class imbalance.

Benchmarks ML and DL models for SQL injection (SQLi) detection on both imbalanced and balanced datasets. Uses consistent preprocessing and evaluation metrics to identify the most effective models and data balancing techniques.

EEG-driven epilepsy detection platform applying ML/DL for focal vs non-focal classification, tackling data imbalance with SMOTE and enabling reliable, faster medical diagnosis.

Fraud detection system using ML and DL models like LightGBM, XGBoost, CatBoost, and Logistic Regression. Optimized with Bayesian tuning to handle imbalanced data, achieving ROC-AUC 0.94 and precision 0.80 for real-time fraud detection in banking systems.

Intro to Machine Learning assignment with step-by-step solutions. Covers basics of AI, ML, DL, and Data Science, types of ML, handling outliers and missing values, encoding techniques, scaling, imbalanced datasets, and Python implementations with sklearn, pandas, and matplotlib.

This repository contains the Jupyter Notebook for my CSCA 5642 Final Project: Fraud Detection using Deep Learning. The project addresses credit card fraud detection on an imbalanced dataset (0.17% frauds), comparing unsupervised (Autoencoder) and supervised (MLP) DL models against a Logistic Regression baseline.

Ranking of drugs that prevent mortality in patients with STEMI. The system is implemented using the Autoencoder/PCA to get the best covariates to 30 days mortality on highly imbalanced data using different ML/DL techniques, and the deepSHAP method is used for model explainability and ranking.

This project demonstrates a comprehensive financial fraud detection system using a variety of ML techniques. It is designed to classify credit card transactions as either legitimate or fraudulent. The solution employs a multi-faceted approach, including supervised, unsupervised, and DL models to handle the challenge of imbalanced fraud data.