Deep Learning Training

Challenge

Deep learning is a complex field that spans numerous specialized and application areas. Our client faced the challenge of finding a tailored training format that practically showcases the entire breadth of the current state of the art to their advanced and experienced users in the field of machine learning using the tools Keras and TensorFlow. Many existing training offerings on the market are geared toward beginners or do not provide an opportunity to engage with real practitioners. The particular challenge of the project was to consolidate the most important topics into a relatively short training duration while developing an effective didactic structure, including internal differentiation.

Approach

For our deep learning bootcamp, we decided on a four-day format in collaboration with the client. This time frame allowed for an in-depth treatment of the central elements of deep learning without overextending the client's time budget.

To design a valuable deep learning course, we first identified the client's most important use cases and compiled state-of-the-art systems in these categories. Behind these various deep learning systems are often similar concepts, which we extracted and developed didactically. Methodologically, we focused on advanced machine learning concepts: the key tools for developing neural networks, computer vision with CNNs, processing sequential data for forecasting or text processing, and working with cloud architectures for the effective deployment of deep learning applications.

Result

As a result, we successfully conducted a four-day deep learning course that brought the client up to date with the discipline through tailored content and tasks. Participants were enabled to decide whether a research problem can be solved with deep learning and, through flexible topic selection and interactive inclusion of their individual questions, were able to actively create methodological value for existing research projects.