Exponential Separations Between Classical and Quantum Learners - IPAM at UCLA

Explore a lecture on exponential separations between classical and quantum learners presented by Vedran Dunjko of Leiden University at IPAM's Mathematical Aspects of Quantum Learning Workshop. Delve into the key challenges of quantum machine learning, focusing on identifying learning problems where quantum algorithms demonstrate provable exponential advantages over classical counterparts. Examine previous examples of quantum learning advantages, their reliance on cryptographic methods, and the intuition that learning separations should be most apparent with data from quantum sources. Discover new results showing various types of advantages obtained from complex quantum physical systems. Investigate how complexity theoretic arguments can prove learning separations in restricted settings where quantum capabilities are only assumed in the training stage. Gain insights into the latest research in this field, referencing works on exponential separations and shadows of quantum machine learning.