SDS 446: Getting Started in Machine Learning — with Jon Krohn

Kernel Tricks

Understanding the kernel trick in support vector machines (SVMs) presents a significant challenge in machine learning. Jon Krohn shares that while executing SVM methods using high-level Python libraries like scikit-learn is straightforward, the intricacies of how the kernel trick functions under the hood can be elusive. This complexity often leads to a gap between practical application and theoretical understanding.

So while it's trivially easy for me to execute support vector machine methods with high level python code saying scikit learn, I've forgotten the details of how the underlying kernel trick works to enable support vector machines to work under the covers.

--- Jon Krohn

This highlights the importance of bridging the gap between using tools and comprehending their underlying mechanisms 1.

   

Reinforcement Learning

Deep reinforcement learning is another area where theoretical understanding can lag behind practical application. Jon Krohn acknowledges the effectiveness of algorithms like deep Q learning for complex tasks, such as video game playing, yet admits to struggling with visualizing the Q learning table updates. This ongoing learning process is what makes the field of machine learning perpetually exciting.

I appreciate how well the deep Q learning algorithm works for complex sequential tasks like playing video games, but I cannot picture the underlying Q learning table updates in my mind yet.

--- Jon Krohn

Jon encourages continuous learning and engagement with the community to tackle these challenges 1 2.