Improved Methods to Inspect Deep Learning Models Will Take Centre Stage this February
Webinar at the World Summit AI will include tips on how to optimize AI projects by inspecting the inner workings of artificial neural networks.
The time is right to learn new tricks of the trade for applied machine learning, so be sure to attend this upcoming webinar made for seasoned data scientists and engineers. Mark your calendars for February 18 at 10:00 am (EST); leaders in data science and software development at Zetane will present the new software tool, the Zetane Viewer, and how we leverage its unique capacities to inspect deep learning models as a means to streamline optimization tasks. While we wait for February to arrive, consider trying the Viewer for yourself — it’s free and available for download now. You can register for the free webinar at this link.
While you’re at it, be sure to check out the other webinars about enterprise AI happening before and after our event on the World Summit AI website.
Here is an abstract for our talk. We look forward to seeing you at the event and hope you can bring a friend.
Why Seeing is Believing… and Understanding: Navigating Inside Complex Neural Networks to Accelerate Discovery by Reducing Guesswork
The microscope, telescope and brain imaging equipment are examples of tools that provide new ways for us to inspect complex systems. Here we present new software tools to do likewise with complex artificial neural networks. New tools require that we develop new techniques to represent and inspect machine learning models, which will also be a topic of discussion. Data scientists from Zetane Systems will demonstrate their development of new digital tools to visually inspect machine learning model architecture and tensors for the purpose of accelerating discovery and innovation in AI. Their presentation will include a case study of model development where they implemented U-Net with medical images. This case study will highlight better optimization strategies provided by inspecting the inner structures of the U-Net model.