Machine learning conference in Munich
It would be difficult to imagine a machine learning conference in Munich without some b.telligent presence – and the b.telligent data science team certainly wasn't going to miss out on a three-day machine learning conference in Munich and an opportunity to share ideas and opinions with like-minded people and learn the latest industry news and trends.
Dr. Sebastian Petry summarized the most important insights as follows:
Deep learning and artificial intelligence (AI) have become an indispensable part of the machine learning process. There was hardly a talk that did not mention this. Many of the talks expressed the view that, despite its huge potential, the discipline is still at a very early stage. Deep learning projects are, after all, still research projects. We were pleased to see our own views on this matter confirmed, and it was also good to see confirmation of our experience with Tensorflow.
How do I measure intelligence with AI? A methodological framework was presented that attempted to answer this fascinating question. It is a question that I intend to explore further, because I see good opportunities for using this framework to evaluate bots and other AIs.
Data science and software development are moving closer together. As smart algorithms continue to become more closely linked to products, data scientists are becoming more closely involved with software development.
I am curious to discover how the concept of “data scientist” will develop in the future. There are two possibilities.
- The technical data scientist: a data scientist who understands the entire set of methods and whose primary interest lies in implementing and optimizing these algorithms in a product.
Serious software development skills are required to do this, even in Java, which is not a typical language for data scientists. The business problem and business case have already been worked out and judged to be promising.
- The business data scientist: a data scientist who is heavily involved in using modern data science methodologies to solve business problems, but is less interested in the actual implementation process. His main focus is the pre-implementation stage, although he has greater involvement with PoC and implementation in less complex environments.
It remains to be seen whether the job market can provide enough data scientists capable of straddling both areas, but I think that the division I have just described makes sense.
Be that as it may, the three-day machine learning conference in Munich proved to be a successful event with an interesting lecture program!