# TBA

Machine learning is a powerful tool, yet we often do not know how well a learning algorithm might perform on any given task. One standard approach to bound the accuracy of a learning algorithm is to reduce the learning task to hypothesis testing, which involves guessing the value of an unknown parameter given a set of observations. Fano's inequality then states that a large amount of correlation between the learner's observations and the set of unknown parameters is a necessary condition for success. In this talk we will discuss how such a condition is also sufficient for succeeding at some learning task, thereby providing a purely information-theoretic guarantee for learning. We will then discuss ongoing efforts to generalize these results to the setting of quantum information theory. We consider a situation where the unknown parameters of the learning task are used to prepare an entangled bipartite state, and the learner's success is measured by their ability to recover quantum correlations between registers via local operations. Our analysis extends learning theory to a more general setting, providing rigorous bounds and guarantees for applying machine learning to a new category of quantum data.

---

Zoom link https://pitp.zoom.us/j/91432159000?pwd=WnlwSGVxcEZkaGFMMHZtVUk5RFJWdz09