Neowin News Feed for: Textfooler Neowin News Feed for: Textfooler en-us Neowin Ignition News (Managing Editor) (Neowin Developers) 5 TextFooler tricks venerable NLP models like BERT into making wrong predictions <div style="float:left;margin-right:10px;"><img src="" alt="" /></div>Researchers at MIT have created a framework—TextFooler—that brought down the prediction accuracy of certain NLP models from 90% down to under 20% by simply using synonyms in place of certain words. <a href="">Read more...</a> (Ather Fawaz) Sat, 08 Feb 2020 07:08:01 +0000