As the area is evolving very quickly, I recommend adding dates to course material, especially, to interviews with experts. It helps us better contextualize emerging trends like Hinton’s capsule, Goodfellow’s GAN, Abbeel’s RL, etc. I think even regular course material (not interviews) could benefit from date annotation, e.g., the old version of ML course mentions 60/20/20 partition of dataset, but newer courses qualify this recommendation for big data.
1 Like
It’s a good suggestion, but I have not heard of any plans to update the DLS courses at this point. For the record, DLS C1 - C4 were published in the latter half of 2017 and C5 was published in early 2018. They were all updated in April 2021 to upgrade from TensorFlow v1.x to TensorFlow 2.x with Eager Mode. At that time, some additional material was added to the ConvNets (C4) and Sequence Models (C5) courses to cover a few additional topics.
When Professor Ng talks about datasets and tuning models in DLS Course 2 and Course 3 he definitely uses more up-to-date thinking for big data than the 60/20/20 partitioning that you mention.
2 Likes
Thank you for your detailed response, and sharing the dates.