"As the machine learning (ML) community continues to accumulate years of experience with live systems, a wide-spread and uncomfortable trend has emerged: developing and deploying ML systems is relatively fast and cheap, but maintaining them over time is difficult and expensive".
(text and image source - Sculley, et al., 2015)
In this talk, we are going to take a look at some problems that arise when we are ready to bring NLP models to production.
- Large models perform well, but how to use them at scale?
- How to monitor their performance on new, streaming data?
- How can we simplify the infrastructure required by these models?
Angelo Basile (http://angelobasile.it), a research scientist at Symanto Research GmbH., will discuss some solutions to these problems.
The talk will be given in English.
Как нас найти:
Организатор: Natural Language Processing group Belarus
Official group of Natural Language Processing and Computational Linguistics in Belarus [http://nlproc.by/] The group for those who are interested in computational linguistics and natural language processing (NLProc). If you heard about part-of-speech tagging, dependency parsing, syntax tree, speech recognition, search engines, voice generation, information extraction, so the group is the one you've better follow. GATE, Apache OpenNLP, Apache UIMA, Apache Stanbol, RDF and OWL, Apache Jena, Lingpipe, Stanford NLProc Software Packages and much more are buzz words in NLProc area. “Human knowledge is expressed in language. So computational linguistics is very important.”–Mark Steedman, ACL Presidential Address (2007)