(We published an early draft pre-print a few months ago, but this final ACL 2018 version has many more experiments, citations, etc.)
-
-
এই থ্রেডটি দেখানধন্যবাদ। আপনার সময়রেখাকে আরো ভালো করে তুলতে টুইটার এটিকে ব্যবহার করবে। পূর্বাবস্থায়পূর্বাবস্থায়
-
-
-
I’m not a NLP guy but this is neat.
-
Nor was I until very recently. It's a good time to become an NLP guy - and extremely important in radiology since basically all the useful labels are in natural(ish) language for now. And the current standard approaches for NLP in radiology are pretty awful
-
True about use NLP in radiology: http://blog.qure.ai/notes/teaching-machines-read-radiology-reports … However, I have to disagree with all the useful labels being there. The amount of intra-radiologists variance means you need either a lot more data or better reference labels.
-
If that's not where all the useful labels are now, then where are they? (Note that I'm not saying we wouldn't like to have better structured labels - I'm just saying that we don't currently have them, on the whole)
-
Well were possible surely you would prefer biopsy - I realised after tweeting that my pov was narrowly scoped to certain cases were biopsies can be taken. I agree with structured labels being nicer to work with
কথা-বার্তা শেষ
নতুন কথা-বার্তা -
-
-
did you created data set proeprly per https://arxiv.org/pdf/1801.06261.pdf … Investigating the Working of Text Classifiers Devendra Singh Sachan∗ Manzil Zaheer∗ Ruslan Salakhutdinov
-
yes it is strange paper, they found that other wrong/trick with a performance by wrong splitting, but by themselves do not share code to make sure their result is correct and even do not answer on emails
@rsalakhu never Prof. Salakhutdinov never his co-authors@sachan_devendra
কথা-বার্তা শেষ
নতুন কথা-বার্তা -
-
-
where is link to paper and code, pls
-
In the beginning of the post. Or here https://arxiv.org/abs/1801.06146
-
ok likn to code is https://github.com/fastai/fastai , but which file to run, may you help to find exactly which fiels/project to run for this paper
কথা-বার্তা শেষ
নতুন কথা-বার্তা -
-
-
Very interesting! Would this technique transfer to medical text which uses a language onto itself?
ধন্যবাদ। আপনার সময়রেখাকে আরো ভালো করে তুলতে টুইটার এটিকে ব্যবহার করবে। পূর্বাবস্থায়পূর্বাবস্থায়
-
লোড হতে বেশ কিছুক্ষণ সময় নিচ্ছে।
টুইটার তার ক্ষমতার বাইরে চলে গেছে বা কোনো সাময়িক সমস্যার সম্মুখীন হয়েছে আবার চেষ্টা করুন বা আরও তথ্যের জন্য টুইটারের স্থিতি দেখুন।