Alexander Kayed

Passionate on all .net things: From machine learning to game development to language design, Alex currently working at Noser Engineering AG as a Senior Software Engineer & Architect in a wide range of projects. In his free time, Alex is climbing, hiking or working on open-source:

Bringing the Power of Deep Learning to .NET with ONNX and ML.NET

With the advent of ChatGPT, everyone is talking about AI. And it is destined to become even more embedded into our daily lives. However, while .NET is usually not associated with machine learning, ML.NET is a powerful tool within our ecosystem and allows us to use our existing C#-skills to train and use machine learning models. The talk will start with an introductory look at ML.NET. Getting started with AutoML and Model Builder is easy and training such small models is feasible for everyone. However, training state-of-the-art large models is near impossible and thus, we can integrate existing, pretrained deep learning models. In the talk, we will use ONNX, an open format to exchange deep neural networks between tools and frameworks, to integrate a pretrained BERT model into a .NET application. BERT is a transformer-based natural language model, similar to the famous GPT-models, and will answer questions in natural language for us.