报告题目：How will Deep Learning be 30 Years from Now?
Adriano Tavares is an Associate Professor at University of Minho, Portugal. He holds a PhD in Industrial Electronics from University of Minho, a Master of Science in Information Technology and an undergraduate degree in Informatics both from University of Coimbra. His main research interests are embedded systems modelling and design, system software design, system-on-chip design/prototyping on FPGA and engineering education. He is currently teaching embedded and real-time systems modelling and design, compiler design, system-on-chip design, real-time operating system design, and advanced computer architectures courses at University of Minho. He published more than 100 book chapters and papers on international conferences and journals related to embedded software and hardware design and two books on microcontroller and C/C++ programming. For the last decade he has been visiting several Asian Universities (e.g., Jilin University-China, Amir Kabir University-Iran, AIT-Thailand, PSUT-Jordan and RUPP-Cambodia) under Erasmus Program. Currently, he is an Invited Professor and researcher in cognitive architecture at University of Trento, Italy.
Deep learning (DL), despite its remarkable successes, is a young field. It is often the case that young fields start in a very ad-hoc manner and later, the mature field is understood very differently than it was understood by its early practitioners. Right now, DL is held together by an extremely successful tool, which does not seem fundamental. It is something we have stumbled on, with seemingly arbitrary details that change regularly. As a field, we do not yet have some shared understanding. In fact, the field has several competing narratives! It seems quite likely DL is in this ad-hoc state.