Fecha de inicio
Abstract: We are witnessing unprecedented growth with new algorithms and applications in Machine Learning and AI, especially with the advent of deep learning and major applications to speech and NLP. However, ML is limited by the quantity and quality of training data, including expert "labels" or "answers" to training problems. Two general methods are presented for overcoming label sparsity, proactive learning and transfer/multitask learning, including their applications to inducing protein-interaction networks and especially to NLP tasks, such as intelligent chatbots and NLP for rare languages.
(la charla se hará en castellano)
Bio:
https://www.cs.cmu.edu/~jgc/
https://en.wikipedia.org/wiki/Jaime_Carbonell