You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
Ablation studies are crucial for deep learning research -- can't stress this enough. Understanding causality in your system is the most straightforward way to generate reliable knowledge (the goal of any research). And ablation is a very low-effort way to look into causality.
If you take any complicated deep learning experimental setup, chances are you can remove a few modules (or replace some trained features with random ones) with no loss of performance. Get rid of the noise in the research process: do ablation studies.
Can't fully understand your system? Many moving parts? Want to make sure the reason it's working is really related to your hypothesis? Try removing stuff. Spend at least ~10% of your experimentation time on an honest effort to disprove your thesis.
Sorry francois what's an ablation study?
Ablation studies are where you subtract from your ML model as a way of understanding the contributions of each component. Sometimes that's extreme (visual caption: removing everything but the language modeling component, no vision) and other times simple (no residual connection).
Agreed! Our #ACL2018 paper (https://arxiv.org/abs/1805.05492 ) does something very similar and shows that state-of-the-art questions answering models largely ignore important parts of the question, leading to highly efficient construction of adversarial attacks.
When there are many features, brute-force ablation doesn’t help. We use a method called “Integrated Gradients” (https://arxiv.org/abs/1703.01365 ) that helps us determine which features are worth ablating.
I can't agree more!
Ablation might be a step forward but ablation is not causality: http://philsci-archive.pitt.edu/12543/
Tu pragmatismo y honestidad se vé muy poco en este ámbito.
@threader_app compile
Hello John, the whole thread from @fchollet is compiled now. Read it here:https://threader.app/thread/1012721582148550662 …
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
