Deep Learning State of the Art (2020)

This podcast takes listeners on an exciting journey through the major advancements in deep learning from 2017 to 2019, while also offering insights into what we can expect in 2020. The speaker highlights significant developments in deep learning frameworks like TensorFlow and PyTorch, breakthroughs in natural language processing with transformers and BERT, and progress in reinforcement learning through systems like AlphaStar and OpenAI Five. Notable moments include the 2019 Turing Award for deep learning and an ongoing discussion about the limits of current deep learning technologies. The episode wraps up by stressing the importance of further research in areas such as common sense reasoning and active learning, as well as the need for greater transparency and ethical approaches in AI, especially regarding recommendation systems. Overall, listeners come away with a deeper understanding of recent achievements and future challenges in deep learning, underscoring the importance of interdisciplinary collaboration and responsible innovation.