Miles Stoudenmire - Flatiron Institute Tensor networks are factorizations of very high order tensors into contracted products of small tensors which can offer exponential gains in memory and computing time. Because of their power and flexibility, tensor networks are finding applications in ever wider settings, including machine learning. Within machine learning, tensor networks define a class of model functions which offer many of the theoretical benefits of kernel methods and scalability similar to neural networks. The wide variety of optimization algorithms and the theoretical underpinnings available for tensor networks point to future opportunities for insights such as matching model architectures to classes of data. I will discuss these opportunities for machine learning research, and highlight exciting recent applications.