Archives: Tweets
-
Props to Ganchev & Das for using Wizard of Oz…
Props to Ganchev & Das for using Wizard of Oz Bulgarian/English example bitext, tho #emnlp2013 #keepinitweird
-
Ganchev & Das plenary: cross-lingual discrim t…
Ganchev & Das plenary: cross-lingual discrim training for sequence models. [Confession: trouble keeping up w optimization math] #emnlp2013
-
Just got thanked for http://t.co/t5D0E0ui64 by Dja…
Just got thanked for http://t.co/t5D0E0ui64 by Djamé Seddah, sitting next to me. Knew me “from twitter”. (“sorry,” I said.) #NLP #humblebrag
-
“graduate student ascent” ha #NLP #academia #gall…
“graduate student ascent” ha #NLP #academia #gallowshumor #optimization
-
Spitkovsky: “It’s enough we’re using graduate stud…
Spitkovsky: “It’s enough we’re using graduate student ascent on this [unsupervised learning]; I don’t want to look at the data.” #emnlp2013
-
Spitk: “engineer builds a bridge, can tell you [wh…
Spitk: “engineer builds a bridge, can tell you [what load it bears]; I cannot tell you if this will fall down or when.” #emnlp2013
-
Klein to Spitkovsky: restarts & parallelism? S…
Klein to Spitkovsky: restarts & parallelism? Spitk: grounded systems can be parallelized; whole thing could be random restarted #emnlp2013
-
Elsner to Spitkovsky: don’t your objectives get ti…
Elsner to Spitkovsky: don’t your objectives get tied in here? Spitk: multiple objectives could be a *useful* alternator for lateen EM
-
Spitkovsky et al. working title (tx Jurafsky) was…
Spitkovsky et al. working title (tx Jurafsky) was “the power of forgetting and starting over” – relevant for neurosci view too #emnlp2013
-
Spitkovsky et al.:build your (EM-like) training fo…
Spitkovsky et al.:build your (EM-like) training fork/join inductors. Structured thinking about inductors and search network #emnlp2013