The author is curious about whether the theoretical knowledge is sufficient for practical applications, particularly in smaller companies that want to apply established methodologies to their own datasets. They are seeking insights on whether this approach is commonly used in practice and if not, they are interested in understanding the reasons behind it.
Key takeaways:
- The author is an undergraduate student interested in deep learning and is curious about the practical application of deep learning theory.
- They have observed that a lot of hyperparameter selection decisions in deep learning come from the engineer's intuition and past experience.
- The author believes it might be possible to develop a rigorous and procedural way to determine hyperparameters and possible architectures based on certain properties of the dataset and intended task.
- They are seeking opinions on whether the theory of deep learning is good enough for practical applications, especially in small companies, and if people actually apply this theory in practice.