BEGIN:VCALENDAR VERSION:2.0 PRODID:ILLC Website X-WR-TIMEZONE:Europe/Amsterdam BEGIN:VTIMEZONE TZID:Europe/Amsterdam X-LIC-LOCATION:Europe/Amsterdam BEGIN:DAYLIGHT TZOFFSETFROM:+0100 TZOFFSETTO:+0200 TZNAME:CEST DTSTART:19700329T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:+0200 TZOFFSETTO:+0100 TZNAME:CET DTSTART:19701025T030000 RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU END:STANDARD END:VTIMEZONE BEGIN:VEVENT UID:/NewsandEvents/Archives/2022/newsitem/13971/13 -December-2022-Computational-Linguistics-Seminar-T ejaswini-Deoskar DTSTAMP:20221208T163731 SUMMARY:Computational Linguistics Seminar, Tejaswi ni Deoskar ATTENDEE;ROLE=Speaker:Tejaswini Deoskar (Utrecht U niversity) DTSTART;TZID=Europe/Amsterdam:20221213T160000 LOCATION:Room L3.36 at LAB42, Amsterdam Science Pa rk, plus live streaming on Zoom. DESCRIPTION:In this talk, I will discuss ongoing r esearch on several topics, on the general theme of generalisation in natural language models. First, I will talk about the generalisation problem in a nalytically complex syntactic parsers, where it is necessary to go beyond supervised models, for ins tance for parsing out-of-domain data or low-resour ce languages; specifically I will present recent r esults on constructing complex category types (in CCG or other categorial grammars) that are unseen in the training data, on-the-fly. Second, I will d iscuss a use-case for a syntactic parser applied t o a new domain: detecting syntactic markers of “ag ency” in language use. Loss of agency is correlate d to psychological conditions like depression or f atigue syndrome, and often expressed in the langua ge produced by patients (e.g. in excessive use of passives). Automatic detection of such markers can help medical professionals intervene and predict recovery in online treatments. Third, I will discu ss recent research on incorporating image-external knowledge for “contextualised” image-captioning: here we develop a generalisable system that can id entify broad-coverage external knowledge relevant to an image. The system can generate informative a s well as factually-correct captions, and be appli ed to various image-language scenarios. X-ALT-DESC;FMTTYPE=text/html:\n
In this talk, I will discuss ongoing research on several topics, on the general theme of generalisation in natural language models. First, I will talk about the gen eralisation problem in analytically complex syntac tic parsers, where it is necessary to go beyond su pervised models, for instance for parsing out-of-d omain data or low-resource languages; specifically I will present recent results on constructing com plex category types (in CCG or other categorial gr ammars) that are unseen in the training data, on-t he-fly. Second, I will discuss a use-case for a sy ntactic parser applied to a new domain: detecting syntactic markers of “agency” in language use. Los s of agency is correlated to psychological conditi ons like depression or fatigue syndrome, and often expressed in the language produced by patients (e .g. in excessive use of passives). Automatic detec tion of such markers can help medical professional s intervene and predict recovery in online treatme nts. Third, I will discuss recent research on inco rporating image-external knowledge for “contextual ised” image-captioning: here we develop a generali sable system that can identify broad-coverage exte rnal knowledge relevant to an image. The system ca n generate informative as well as factually-correc t captions, and be applied to various image-langua ge scenarios.
URL:https://projects.illc.uva.nl/LaCo/CLS/ CONTACT:Alina Leidinger at a.j.leidinger at uva.nl END:VEVENT END:VCALENDAR