Day 2 Technical Track - One Universal Language for All Machine Learning Models

Anna Jung, ML Open Source Engineer, VMware

Deep Learning is used across multiple industries to solve difficult problems like natural language processing and computer vision. With the help of deep learning frameworks like Tensorflow, PyTorch, and Keras, Machine Learning engineers can quickly build complex models like a virtual assistant to recognize speech and formulate a response without being an expert in machine learning. However, one of the biggest challenges in the field is choosing the right framework for the use case. With no interoperability across frameworks, engineers often find themselves locked into a specific vendor limited to the initially selected framework's capabilities.

With Open Neural Network Exchange (ONNX), Machine Learning engineers no longer have to worry about limitations. They can pick and choose the tools they need to build their model by leveraging the ONNX open format. Using ONNX, all machine learning models become reusable and portable, enabling engineers to do more by moving from one tool to another and training and inferencing using different tools.

So, what exactly is ONNX? Simply put, ONNX is a universal language for all machine learning models that enable engineers to make better, faster, smarter models without being locked into an ecosystem. In this session, I will explore in detail what ONNX is, why it matters, and how you can get involved with the project.

For more information about this session and other sessions at the 2021 Women Transforming Technology Conference, please visit the website:
https://www.womentransformingtechnology.com/