minsoo9506 / catchMinor-research

research about imbalanced learning & anomaly detection (tabular, time series)

• research about Imbalanced Learning & Anomaly Detection, Outlier Analysis
  • tabular, time series, graph

📖 Index

• Practice
• Project
• Paper Read
  • Imbalanced Learning
    • Survey
    • Perfomance Measure
    • Cost-sensitive
    • Sampling
    • Ensemble Learning
    • Feature Selection
    • Imbalanced Classification with Multiple Classes
  • Anomaly Detection, Outlier Analysis
    • Outlier Analysis (2017) - Charu C. Aggarwal
    • Survey
    • Learning feature representations of normality
    • Time Series and Streaming Anomaly Detection
• Other References
  • Industry
  • Article

Practice

• Dacon 신용카드 사용자 연체 예측 AI 경진대회 code
  • task: tabular, multiple classes classification(3 classes), imbalance
  • method: OVO + Oversampling, Probability Calibration, MetaCost
• Kaggle Credit Card Fraud Detection code
  • task: tabular, binary classification, imbalance
  • method: SMOTE, Unsupervised PCA based algorithm

Project

• 네트워크임베딩 대학원수업 기말 프로젝트 (Anomaly Detection with Graph Embedding Ensemble) pdf
  • task: tabular data, graph embedding, anomaly detection
  • method: Node2Vec, PCA, Mahalanobis, LOF, Random Forest
• 석사 졸업 논문 (Anomaly Detection with Adaptive-AutoEncoder Ensemble) repository
  • task: tabular data, ensemble, anomaly detection
  • method: AutoEncoder
• 모델 구현 (라이브러리화) repository

Paper Read

Imbalanced Learning

Survey

Survey

• Learning From Imbalanced Data: open challenges and future directions (survey article 2016) paper

Perfomance Measure

Perfomance Measure

• The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets paper
• The Relationship Between Precision-Recall and ROC Curves paper
• Predicting Good Probabilities With Supervised Learning paper
• Properties and benefits of calibrated classifiers paper
• The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets paper

Cost-sensitive

Cost-sensitive

• An optimized cost-sensitive SVM for imbalanced data learning paper
• Metacost : a general method for making classifiers cost-sensitive (KDD 99) paper
• The influence of class imbalance on cost-sensitive learning (IEEE 2006) paper
• Learning and Making Decisions When Costs and Probabilities are Both Unknown (2001) paper

Sampling

Sampling

• SMOTE (2002) paper
• SMOTE for learning from imbalanced data : progress and challenges (2018) paper
• Influence of minority class instance types on SMOTE imbalanced data oversampling paper
• Calibrating Probability with Undersampling for Unbalanced Classification (2015) paper
• A Study of the Behavior of Several Methods for Balancing Machine Learning Training Data paper
• Dynamic Sampling in Convolutional Neural Networks for Imbalanced Data Classification paper review

Ensemble Learning

Ensemble Learning

• Self-paced Ensemble for Highly Imbalanced Massive Data Classification (2020) paper

Feature Selection

Feature Selection

• Ensemble-based wrapper methods for feature selection and class imbalance learning (2010) paper
• A comparative study of iterative and non-iterative feature selection techniques for software defect prediction

Imbalanced Classification with Multiple Classes

Imbalanced Classification with Multiple Classes

• Imbalanced Classification with Multiple Classes
  • Decomposition-Based Approaches
  • Ad-hoc Approaches

Anomaly Detection, Outlier Analysis

Outlier Analysis (2017) - Charu C. Aggarwal

Outlier Analysis (2017) - Charu C. Aggarwal

Suvey

Survey

• Deep Learning for Anomaly Detection A Review (2020) paper review
• Autoencoders (2020) paper

Learning feature representation of normality

Learning feature representations of normality

• Outlier Detection with AutoEncoder Ensemble (2017) paper
• Auto-Encoding Variational Bayes (2014) paper review code
• Deep Variational Information Bottleneck (ICLR 2017) paper review
• Extracting and Composing Robust Features with Denoising Autoencoders (2008) paper
• Generatice Adversarial Nets (NIPS 2014) paper review code
• Least Squares Generative Adversarial Networks (2016) paper review
• Adversarial Autoencoders (2016) paper review
• Generative Probabilistic Novelty Detection with Adversarial Autoencoders (NIPS 2018) paper
• Deep Autoencoding Gaussian Mixture Model For Unsupervised Anomaly Detection (ICLR 2018) paper review
• Anomaly Detection with Robust Deep Autoencoders (KDD 2017) paper

Time Series and Streaming Anomaly Detection

Time Series and Streaming Anomaly Detection

• Anomaly Detection In Univariate Time-Series : A Survey on the state-of-the-art paper
• USAD : UnSupervised Anomaly Detection on multivariate time series (KDD2020) paper review
• Variational Attention for Sequence-to-Sequence Models (2017) paper
• A Multimodal Anomaly Detector for Robot-Assisted Feeding Using an LSTM-based Variational Autoencoder (2017) paper
• Outlier Detection for Time Series with Recurrent Autoencoder Ensembles (2019) paper
• Robust Anomaly Detection for Multivariate time series through Stochastic Recurrent Neural Network (KKD 2019) paper
• Time Series Anomaly Detection with Multiresolution Ensemble Decoding (AAAI 2021) paper
• An Improved Arima-Based Traffic Anomaly Detection Algorithm for Wireless Sensor Networks (2016) paper
• Time-Series Anomaly Detection Service at Microsoft (2019) paper
• Time Series Anomaly Detection Using Convolutional Neural Networks and Transfer Learning (2019) paper code
• Abuse and Fraud Detection in Streaming Services Using Heuristic-Aware Machine Learning (arxiv, 2022, Netflix) paper
• Are Transformers Effective for Time Series Forecasting?, 2022 paper

Other References

Industry

• (Netflix) Machine Learning for Fraud Detection in Streaming Services
• Fraud Detection on Blockchain based on Machine learning (medium article)
• (Pinterest) Warden: Real Time Anomaly Detection at Pinterest
• (배달의 민족) 배민 앱 리뷰 품질을 향상시킨 방법은? 머신 러닝 X 네트워크 탐지 모델 도입
• Fighting Spam using Clustering and Automated Rule Creation
• (Lyft) Full-Spectrum ML Model Monitoring at Lyft
• (Lyft) Building a large scale unsupervised model anomaly detection system — Part 1
• (Lyft) Building a large scale unsupervised model anomaly detection system — Part 2

Article

• Unsupervised and semi-supervised anomaly detection with data-centric ML, google blog