分野別セミナー

The area of neural networks has undergone revolutionary changes in the past 4 years. This has resulted in moving neural networks from the fringes of the machine learning community to center stage. Recently researchers have discovered effective ways of training neural networks that are 10 layers deep that contain billions of parameters. These deep neural networks have produced outstanding results in the area of image recognition, speech recognition, natural language processing, multi-modal learning, reinforcement learning, etc. For example in the area of image recognition the state-of-the-art deep learning system can achieve an error rate under 5%, which is more accurate than most humans can achieve.


In this talk I will present some recent applications of deep learning to solve very interesting problems. I will also present some intuitive explanations of how the algorithms work.