We all face the never-ending race to collect more labeled data for training our models. However, it’s not all about data volume - but also its quality. By collecting labeled data wisely, you can achieve equivalent or better performance levels with fewer data samples - thus saving time and money. How can we do it? By using Active Learning!
In this talk, Sivan will share her work in the field of Active Learning. She will introduce methods she used for collecting data wisely and share the considerations behind them. She will describe the research process and its results. Most importantly, she will introduce ways to validate those results, which can be useful to you and your work.