Products & AI
Smart labeling classification datasets mixing machine learning with deep learning, using Microsoft ML.NET
Daniel Costea - Senior Software Developer @ EU Agency
Live from Studio room
11th November, 11:30-12:30
Supervised learning relies on labeled datasets. Since most data are not labeled, the labeling process is very painful, especially when it's done by humans. Why not use computer vision to delegate this burden? The solutions you can find on the market are for labeling images datasets, but I couldn't find labeling numerical datasets that are typical for machine learning. This solution will mix together the power of machine learning multiclassification with the power of transfer learning (by image classification). In this talk, I will demo how I have used image classification to label data from an IoT device (Raspberry Pi) running .NET 5 on Linux.