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/2008/newsitem/2663/15- December-2008-From-Perception-to-Symbols-Amitabha- Mukerjee DTSTAMP:20081211T000000 SUMMARY:From Perception to Symbols, Amitabha Muker jee ATTENDEE;ROLE=Speaker:Amitabha Mukerjee (IIT Kanpu r, India) DTSTART;TZID=Europe/Amsterdam:20081215T140000 DTEND;TZID=Europe/Amsterdam:20081215T150000 LOCATION:Room P.015A, Euclides Building, Plantage Muidergracht 24, Amsterdam DESCRIPTION:Life is... a trap for logicians. It lo oks just a little more mathematical and regular th an it is; its exactitude is obvious, but its inexa ctitude is hidden; its wildness lies in wait. - G . K. Chesterton, The Paradoxes of Christianity W hile it is clear that there are a lot of regularit ies in the world, whether these can be handled by formal, symbolic processes is a question that is b eing widely debated today. In this work, we presen t computational evidence for the process by which symbols may arise from perceptual input (2D and 3D videos). Based on cognitive psychology models of pre-linguistic "perceptual schemas", we use unsupe rvised clustering on an intuitive feature space fo r object shapes, relative positions, and actions, we show how concepts corresponding to nouns, prepo sitions and verbs can be acquired. Finally, we sho w how these "prior" schemas can now be associated with linguistic commentary to pick out the words t hat best correspond to the concepts in the target language. These associations, with a semantic pole (the grounded perceptual schema) linked to a phon ological pole (the word learned) constitute our el ementary symbols. One of the primary differences w ith formal models of semantics is that the meaning associations for these symbols are plastic, and c an be altered by subsequent experience. Issues reg arding the composition of symbols indicate a prefe rence for frequently occurring data, which is also typical of linguistic usage. X-ALT-DESC;FMTTYPE=text/html:\n\n Life is... a trap for logician s. It looks just a little more mathematical\n and regular than it is; its exactitude is obvious, but its inexactitude\n is hidden; its wildnes s lies in wait.
\n - G. K. Chesterton, The Paradoxes of Christianity\n< /cite>\n

\nWhile it is clear that there ar e a lot of regularities in the world,\nwhether the se can be handled by formal, symbolic processes is a question\nthat is being widely debated today. In this work, we present\ncomputational evidence f or the process by which symbols may arise from\npe rceptual input (2D and 3D videos). Based on cogni tive psychology models\nof pre-linguistic "pe rceptual schemas", we use unsupervised cluste ring on\nan intuitive feature space for object sha pes, relative positions, and\nactions, we show how concepts corresponding to nouns, prepositions and \nverbs can be acquired. Finally, we show how thes e "prior" schemas can now\nbe associated with linguistic commentary to pick out the words that best\ncorrespond to the concepts in the targe t language. These associations,\nwith a semantic pole (the grounded perceptual schema) linked to a\ nphonological pole (the word learned) constitute o ur elementary symbols.\nOne of the primary differe nces with formal models of semantics is that the\n meaning associations for these symbols are plastic , and can be altered by\nsubsequent experience. Is sues regarding the composition of symbols\nindicat e a preference for frequently occurring data, whic h is also typical\nof linguistic usage.\n

\n URL:/NewsandEvents/Archives/2008/newsitem/2663/15- December-2008-From-Perception-to-Symbols-Amitabha- Mukerjee END:VEVENT END:VCALENDAR