In this episode I speak with Jon Krohn, author of Deeplearning Illustrated a book that makes deep learning easier to grasp.
We also talk about some important guidelines to take into account whenever you implement a deep learning model. How would one deal with bias in machine learning? John is the lead scientist of a project at the interception of HR and machine learning to match jobs to candidates. We speak about how his team mitigate bias in such scenario. For instance, in order to guarantee fairness with women in IT.
What are the limitations of deep learning technology?
From biological vision to natural language processing, to machine art to the video games of OpenAI. It seems deep learning is the way to go for anything. We discuss some cases in which deep learning is not really the way to go. Other methods, faster and less demanding of data might do a better job.
Deep learning and Reinforcement Learning
John wrote a chapter about reinforcement learning and the integration of deep learning. So far we have seen reinforcement learning doing a great job on narrow problems like the Atari games of OpenAI gym. But very little has been done in real-world applications. John and I discuss some of the limitations of reinforcement learning that prevent it from being widely adopted in every domain.
The impact of deep learning in data science
Despite the high requirements in terms of data and hardware, deep learning is changing the way we do science. We talk about the impact that deep learning is having in the data science community and research field. We also touch another important topic about the role of deep learning in the AI of the future.
Deep learning illustrated: the book
I personally find John’s book an interesting work to have on the shelf. It makes deep learning easier to grasp with the stunning graphics and its effective examples.
Yes! Deep learning is easier when it is illustrated.
You can purchase the book from informit.com/dsathome with code DSATHOME and get 40% off books/eBooks and 60% off video training