Detecting Anomalies with Machine Learning

Übersicht

Engineers are responsible for detecting abnormalities and ensuring reliability in the products they produce. Developments in artificial intelligence (AI) provide new tools for engineers to quicky identify anomalies and anticipate maintenance needs, preventing costly downtime and failures.

Acquiring these skills will enable you to remain competitive and enhance the quality and reliability of your systems.

Lehrplan

• Introduction to Anomaly Detection
Definition and significance of anomaly detection
Overview of applications in engineering and industry
Types of anomalies: point, contextual, and collective
• Machine Learning Fundamentals
Supervised vs unsupervised learning
Overview of classification and clustering techniques
Evaluation metrics: precision, recall, F1 score, and ROC-AUC
• Data Preprocessing for Anomaly Detection
Data collection and data types
Data cleaning and handling missing values
Feature selection and dimensionality reduction
Normalization and standardization
• Unsupervised Methods for Anomaly Detection
Clustering-based approaches: k-means, DBSCAN
Density-based methods: Isolation Forest, Local Outlier Factor
Autoencoders for anomaly detection
• Supervised Methods for Anomaly Detection
Choosing the right labels for anomaly detection
Classification techniques for anomaly detection
Time-series anomaly detection
• Real-Time Anomaly Detection
Streaming data and continuous monitoring
Implementing real-time anomaly detection systems
Performance considerations in real-time systems
• Deploying Anomaly Detection Models
Model deployment strategies
Integrating anomaly detection in maintenance and monitoring workflows
Challenges and considerations in deployment
• Case Studies and Applications
Industrial manufacturing and predictive maintenance
Financial fraud detection
Network security and intrusion detection
• Tools and Platforms for Anomaly Detection
Overview of popular libraries and tools: Scikit-learn, TensorFlow, PyTorch
Cloud-based solutions and services
Building custom solutions with open-source tools
• Final Project
Real-world anomaly detection project
Dataset selection and problem definition
Building, evaluating, and presenting the anomaly detection model

---

Unterrichtet von

Megan Thompson, Kathy Tao, Rohit Ramanathan, Marissa D'Alonzo and Brian Buechel

---

Fachgebiete

Computer Science