[19] Dumitru Erhan - Understanding Deep Architectures and the Effect of Unsupervised Pretraining
The Thesis Review
00:00
Transformations in Deep Learning
This chapter explores the evolution of image recognition technology at Google during the deep learning revolution around 2012. The speaker shares their insights on the transition from traditional computer vision to advanced deep neural networks, emphasizing significant milestones like AlexNet and GoogleNet. The discussion also covers challenges in reinforcement learning, the debate over model-free versus model-based approaches, and the complexities of applying unsupervised learning in practice.
Transcript
Play full episode
