Welcome to your comprehensive guide on the KDD’22 tutorial entitled “Robust Time Series Analysis and Applications: An Industrial Perspective.” This tutorial will not only introduce you to advanced time series analysis but will also provide insightful applications across various industrial fields. So, buckle up as we dive into this enriching experience!

Tutorial Details

• Date: Sunday, 08/14/2022
• Time: 9:00 am – 12:00 pm
• Location: Room 207B, Washington DC Convention Center, USA

Tutorial Abstract

Time series analysis plays a crucial role in various areas like AIOps in cloud computing, AI-powered Business Intelligence in E-commerce, and the Artificial Intelligence of Things (AIoT). However, time series data can often present complex patterns influenced by trends, seasonality, outliers, and noise. As we accumulate significant volumes of time series data, the effective management of such data becomes essential for tasks like forecasting, anomaly detection, and fault cause localization. The tutorial intends to equip you with robust time series models designed to handle these intricate challenges.

Tutorial Structure

This tutorial is structured in a bottom-up framework, ensuring a solid foundation before diving into advanced topics:

• Introduction to Real-world Challenges and Needs for Robustness
• Preliminaries including:
  • Robust Statistics: Robust Regression, M-estimators
  • Signal Processing: Fourier, Wavelet
  • Optimization Algorithms: ADMM, Majorize-Minimization
  • Deep Learning Techniques: RNN, CNN, GNN, Transformer
• Robust Time Series Processing Blocks:
  • Time Series Periodicity Detection
  • Time Series Trend Filtering
  • Time Series Seasonal-Trend Decomposition
  • Time Series Similarity
• Robust Time Series Applications and Practices covering:
  • Forecasting: Tree Model, Deep Ensemble, Transformer, with Case Studies
  • Autoscaling from Forecasting to Decision-Making: Query Modeling and Scaling Decision
  • Anomaly Detection: Decomposition Model, Deep State Space Model, Transformer
  • Fault Cause Localization: Rule Set Learning and Root Cause Analysis
• Further Reading: Selected AI for Time Series papers, tutorials, and surveys can be accessed via GitHub here.

Learning Outcomes

By attending this tutorial, you will gain insights into both traditional statistical methods and state-of-the-art algorithms in time series analysis. You will also learn how to tackle practical problems encountered in industrial applications.

Analogy – Understanding Robust Time Series Analysis

Imagine you are trying to listen to your favorite music in a crowded café. The clatter of dishes, conversations, and the coffee grinder noise represent data noise in time series analysis. Just as you need to discern your favorite song amidst the café’s noise, time series analysis involves extracting meaningful patterns (your song) from complex data (the café noise). Robust time series models function like noise-canceling headphones, filtering out distractions and enabling you to enjoy the music effortlessly. This capability is indispensable in applications where accuracy is critical, such as forecasting and anomaly detection.

Troubleshooting Tips

So what happens if you face challenges while applying robust time series techniques? Here are some suggestions:

• Ensure your data is preprocessed and cleaned before analysis. Noise and outliers can severely impact model performance.
• If your model isn’t performing as expected, check if your choice of algorithms fits the characteristics of your time series.
• Experiment with various parameters and model architectures to achieve better outcomes.
• If you encounter any technical issues or have specific questions, feel free to reach out for help.

For more insights, updates, or to collaborate on AI development projects, stay connected with fxis.ai.

Conclusion

Robust time series analysis is increasingly important across various industries, and this tutorial provides an invaluable opportunity to explore this field comprehensively. With the knowledge gained, you can significantly improve your analytical capabilities and decision-making processes.

At fxis.ai, we believe that such advancements are crucial for the future of AI, as they enable more comprehensive and effective solutions. Our team is continually exploring new methodologies to push the envelope in artificial intelligence, ensuring that our clients benefit from the latest technological innovations.