Faktion Academy.

Machine Learning for sensor data

This course will teach best practices on deploying ML models, model management, data versioning and useful frameworks to use.

Course outline

General concepts 

  • Sampling theorem
  • Time windows and aggregation
  • Data splitting for time series data
  • Common data quality problems and how to solve them
  • Aligning flow by using dynamic time warping
  • Digital Twins 

Pre-processing sensor data time series 

  • Interpolation
  • Noise reduction 
  • Outlier detection
  • Dimensionality reduction
  • Dealing with mixed sample frequencies 

Feature calculation 

  • Why?
  • Autocorrelation
  • Fourier transformations
  • Peak detection
  • ARIMA models

Forecasting  (on demand)

  • Arima models
  • Exponential smoothing models
  • Anomaly detection based on forecasting 

Predictive maintenance (on demand)

  • Predictive maintenance and survival function estimation 
  • Kaplan Meier estimators
  • Cox Proportional Hazard model
  • Aalen Additive Hazard model
  • Time series similarity kernel regression 

Deep Learning for sensor data (on demand)

  • Recurrent architectures
  • Convolutional architectures
  • Anomaly detection based on auto-encoders
  • Reinforcement learning for optimal control
  • Differential evolution
  • Surrogate/Bayesian optimization 

Case studies and exercises will take place per chapter.

The course outline will be tailored to the participant’s wishes and needs. 

Course level

Expert

Prerequisites

  • Python programming at the intermediate level
  • Knowledge of basic Machine Learning concepts like data splitting, classification, overfitting, probabilities, …

Course teachers

Jeroen Boeye, PhD
Head of Sensor Data

Vladimir Dzyuba, PhD
Senior ML Engineer

Course fee

EUR 3.000 excl. VAT

Included in course package

  • Course material
  • Drinks, snacks and lunch
  • Cloud servers for use during training
  • 4h of support and question answering up to 6 months after the course

The Natural Language Processing course was a perfect balance between a summary view of the NLP journey with its modelling approaches over time and some technical details, plus a practical business view on real-world applications.

Claudia Burgard

Senior Data Scientist at Stepstone

Register now – limited seats

Machine Learning for Sensor Data

8-10/09/2020

We will confirm your registration by e-mail

Get in touch

Questions?

Contact us


Contact

Inquiry for your POC

=