Recorded 29 November 2022. Daniel Cremers of Technische Universtitat München presents "Deep Learning: Challenges and Perspectives" at IPAM's Multi-Modal Imaging with Deep Learning and Modeling Workshop. Abstract: Since 2012 deep networks have revolutionized computer vision and many other areas of science. In my presentation, I will focus on a number of challenges in this field: How can we extend the success of deep networks beyond supervised image classification? How can we combine deep nets with classical optimization methods? And how can we extend convolutional networks to non-Euclidean structures? Learn more online at: http://www.ipam.ucla.edu/programs/workshops/workshop-iv-multi-modal-imaging-with-deep-learning-and-modeling/?tab=overview