Machine Learning with small data
10:30 - 11:00
Modern Machine Learning algorithms are notorious for their high compute and large data requirements - often ranging from millions to billions of data points. However, most small to medium organisations don't have access to such volumes of (labeled) data, yet are very keen to put whatever data they collect to good use. We will investigate here a range of emerging techniques that can benefit anyone trying to employ ML with smaller than ideal datasets: transfer learning, zero-shot learning, semi-supervised learning and others.
Machine Learning Engineer and enthusiast, on a mission to use the power of advanced learning algorithms, extract value from data and unleash its full potential to ultimately impact people's lives for the better. Avid reader of new research papers and Kaggle competitor in his spare time.