Distributed TensorFlow and Classification of Time Series Data Using Neural Networks

by Sophia TurolAugust 8, 2016

Learn about the advantages of deep learning technologies for the classification of time series data and the benefits of distributed TensorFlow for model training.

While the conventional methods may lag behind in terms of accuracy, convolutional neural networks come to the rescue. At the recent TensorFlow meetup in Munich, the speakers highlighted the potential of deep learning tools for classifying time series data, as well as the perks of training models using distributed TensorFlow.

Table of Contents

Classifying time series data with TensorFlow

In his session, Andreas Pawlik of NorCom explored the potential of deep learning technologies and TensorFlow for classification of time series data, or timestamps used for sequence of events. He enumerated a few examples for applying this approach:

Light curves in astrophysics
Different shapes (skull, blood cell, butterfly, etc.)
Electrocardiograms in medicine
Protein sequences in genetics
Intruder activity logs in IT security
Sensor data

Andreas underlined that though classification of time series data is a standard problem, still it can be challenging. Furthermore, conventional approaches may be computationally expensive and their accuracy heavily depends on the quality of user input. In contrast to that, deep learning technologies—such as multi-scale convolutional networks—pose a better alternative with high accuracy rate.

Find Andreas’s presentation below.

Intro to distributed TensorFlow

Miha Pelko of NorCom overviewed the distributed version of TensorFlow, which allows for training bigger models faster. He demonstrated different ways of model training parallelization for:

Separate models with different data
Separate models with the same data
Full model with split data
Split model with full-split data

Recently, we have conducted a study exploring the performance of distributed TensorFlow. You can find full Miha’s slides below.

Want details? Watch the video!

Join the meetup group to get informed about the upcoming events.

About the experts

Andreas Pawlik is Data Scientist at NorCom. After graduating from the University of Potsdam with a degree in physics, he proceeded to studying astrophysics and worked as Research Scientist in a number of universities. Andreas is currently working with sensor data, such as time series, images, and IoT. He delivers data-driven solutions using deep learning and natural language processing.

Miha Pelko is Data Scientist at NorCom. He holds a PhD in computational neuroscience from the University of Edinburgh, where he researched information processing in neurons using statistical modeling, data analysis, and dynamic system simulation. At the moment, Miha puts his interest in big data and the use of distributed computing in research and development workflows.