Data Science & Deep Learning
The Digital Transformation is causing companies to generate tremendous quantities of highly varied data. This creates huge requirements in terms of the complexity and scalability of analysis methods ('Data Science at Scale').
In order to fulfil these requirements, we apply proven analysis methods from the worlds of science and academia (among others) to the corporate world. We use custom-tailored algorithms to extract valuable information and knowledge from immense volumes of data. One example of scientific methods being used in a business context is provided by multivariate procedures: the same concepts and algorithms used by CERN to search for the Higgs-Boson particle are now being used to analyse extremely large volumes of web traffic data for ProSiebenSat.1.
Team of Interdisciplinary Experts
For several years now, inovex has employed a very diverse team of highly qualified data scientists. These experts work to make the latest data analysis methods available for use in our customer projects. They bring to the table extremely efficient algorithms from their specialised fields of expertise – including biology, mathematics, physics, linguistics and computer science – and use them to create our wide range of services pertaining to predictive and prescriptive analytics.
Integration of Data Science into Big Data Systems
Close interdisciplinary collaboration with our Big Data experts means that the analysis selection and implementation process focuses on the scalability and performance of these algorithms. Mature concepts for maintaining these analyses and integrating them into productive environments are the keys to our solutions' success.
- Batch- and Stream-Analysis with Apache Spark and Flink
- Ad-hoc access with Apache Zeppelin and Jupyter
- Deep Learning with Google Tensorflow
- Implementation in Python, R, Java or Scala
Focus on Open-Source Technologies
Our Data Science team uses established open-source technologies, like the Apache Spark Machine Learning (ML) library and Flink ML. These technologies facilitate both high-performance batch processing and the complex real-time analysis of data streams. Explorative ad-hoc access is provided by interactive notebooks like Apache Zeppelin or Jupyter, while versatile APIs enable implementation in Python, R, Java or Scala, depending on customer preferences. We also use specialist libraries like lightFM, a Python implementation of efficient recommendation algorithms. This approach enables us to cover all the latest processes – like regression, decision trees, support vector machines and neuronal nets – using our in-house resources.
Data Science at Scale
Our extensive technology stack and our experts' wide range of specialist knowledge allow us to provide our customers with a comprehensive 'Data Science at Scale' offering – from clustering and classification to text mining and forecasts, right through to recommendation and fraud detection.
A specialist discipline at the intersection of Data Science and Artificial Intelligence, Deep Learning, or the use of multi-layered neural networks to create learning systems, is becoming increasingly relevant. Since the success of the AlphaGo deep learning program and the popularity of the TensorFlow program library (both from Google), Deep Learning has become increasingly popular – not just in a university context, but also among corporations. When combined with today's high-performance hardware, Deep Learning offers tremendous potential for raising the quality of image analysis (for use in functions like automated classification and indexing), text processing (for automatic translation, for example) and language processing (for use in functions like voice control) to new heights. In the inovex Lab, we focus very intensively on Deep Learning and its associated disciplines, as we believe that these technologies are the way forward.
Would you like a consultation on this subject?
I look forward to hearing from you!
Head of Data Management & Analytics