Target group:

Researchers, engineers, and innovation managers from manufacturing industries (e.g. automotive, semiconductors, machinery…). Representatives from energy, health, and financial domains might also be interested.

Abstract:

Driven primarily by the ongoing digitization and the Industry 4.0 revolution, time series data has become ubiquitous. Innumerable sensors installed at various stages of an industrial production process are generating data at a high rate. Modern IoT (Internet of Things) architectures support the transfer of collected data over high-bandwidth networks and provide means for high-performance processing on the edge and in the cloud. The capability to collect massive amounts of time series data, paired with modern data analytics and machine learning methods, opens the way to novel applications with the potential of greatly benefiting the efficiency and the resiliency of industrial production processes. Besides manufacturing machinery, time-series data are created en masse in numerous other situations such as human diagnostics in medicine, energy production and consumption, agriculture, financial transaction tracking, personal mobile devices (smartphones), and many others. Clearly, methods for time series analytics have a far broader area of application than the manufacturing industry alone.

In this session, we will present selected research contributions addressing the analytics of time series data. In particular, you will hear about distance measures and seasonality detection, event detection and prediction using deep learning methods, and methods for finding patterns in large time-series databases. To illustrate the benefits arising from the application of these techniques, we will present results from selected past or running projects. Examples include modality detection in application domains such as mobility or precision agriculture, event classification for predictive maintenance, or gathering expert knowledge to improve algorithm performance.

Finally, we wish to learn about your needs and how we could address those needs in an effective manner. What problems are you expecting to solve with time-series analytics in your organization? Did you already gather experience in automating data-oriented tasks with AI algorithms? Which problems (e.g. data availability/security, legal issues, employee acceptance) might prevent you to introduce data analytics in your work processes? Time for discussion will be allocated after each talk. Your input and your feedback will be highly appreciated!

After the event you will know:

 

About bleeding-edge methods for analyzing time-series data, such as

• detection and prediction in time-series data


• deep learning on time series


• feature extraction from time series


• searching in time-series information

You will also learn about success stories and applications, such as

• predictive maintenance in manufacturing


• interactive applications for time-series analytics