Introduction to Positive Unlabeled (PU) Learning | DataHour | Analytics Vidhya скачать в хорошем качестве

Introduction to Positive Unlabeled (PU) Learning | DataHour | Analytics Vidhya

Positive-Unlabeled (PU) learning is a Machine Learning approach to Binary Classification where the training data comprises of positive instances as well as an additional unlabeled data that might contain positive and negative instances in unknown proportions. Positive-unlabeled learning methods aim to incorporate this unique scenario into the learning process, in a way that improves generalization of the learned representations of the positive class, when compared to simply treating all unlabeled instances as purely negative instances, or alternatively discarding them and training a one-class classifier over only the positive samples. In this DataHour, Chandra will explain all about Positive Unlabeled learning including its basics, use and practical applications. Sections 00:00:00 Introduction 00:02:20 Machine Learning: Quick Brush-up 00:04:48 What is Positive Unlabeled Learning 00:06:69 Different Approaches to PU Learning 00:10:52 Hands On PU Learning 00:39:44 Some Real-World Application 00:47:46 Brief Q&A Session Stay on top of your industry by interacting with us on our social channels: Follow us on Instagram:   / analytics_vidhya   Like us on Facebook:   / analyticsvidhya   Follow us on Twitter:   / analyticsvidhya   Follow us on LinkedIn:  / analytics-vidhya