Algoritmide ja andmestruktuuride erikursus - Redigeerimiste ajalugu

Tammet: /* Eksamis tulevad teemad */

2019-12-16T08:59:04Z

Eksamis tulevad teemad

←Vanem redaktsioon		Redaktsioon: 16. detsember 2019, kell 10:59
69. rida:		69. rida:

	Võrdussüsteemid:		Võrdussüsteemid:
	* Eestikeelsest loogikaõpikust lk 366-375		* Eestikeelsest loogikaõpikust lk 366-375 termiteisendussüsteemid: oskus väikesest võrdussüsteemist järeldusi teha ja leida täielik kokkuvoolav võrdussüsteem (kui seda saab ehitada)
	termiteisendussüsteemid:
	oskus väikesest võrdussüsteemist järeldusi
	teha ja leida täielik kokkuvoolav võrdussüsteem
	(kui seda saab ehitada)
	* Geoff lk 67-73		* Geoff lk 67-73

82. rida:		78. rida:
	Clause and literal selection: mis meetodid on enam kasutatavad / lihtsad		Clause and literal selection: mis meetodid on enam kasutatavad / lihtsad
	* Schulz lk 117-118		* Schulz lk 117-118



	== Time and place ==		== Time and place ==

Tammet – 16. detsember 2019, kell 08:58

2019-12-16T08:58:34Z

@@ 40. rida: / 40. rida: @@
 * 13.01.2019 12:00 ja 15:00 ruumis U06A-229
 * 20.01.2019 12:00 ja 15:00 ruumis U02-102
 == Time and place ==

Tammet: /* Esmaspäeval ainult loeng kell 10:00 */

2019-12-15T09:30:36Z

Esmaspäeval ainult loeng kell 10:00

←Vanem redaktsioon		Redaktsioon: 15. detsember 2019, kell 11:30
25. rida:		25. rida:


	== <font color="red">Esmaspäeval ainult loeng kell 10:00</font> ==		== <font color="red">Esmaspäeval ainult loeng kell 10:00, praksivariant kell 16</font> ==

	Esmaspäeval loengus kell 10 plaanime eksami sisu ja vaatame üle praktilised tööd.		Esmaspäeval loengus kell 10 plaanime eksami sisu ja vaatame üle praktilised tööd.

	Kella 14-ne praktikum jääb ära: ~~õppejõul~~ samal ajal intervjuu err-le.		Kella 14-ne praktikum jääb ära: annan samal ajal intervjuu err-le.

			Kui peaks kellelegi sobima, siis hea meelega teeksin praktikumi hiljem, kell 16.

	== Exams ==		== Exams ==

Tammet – 15. detsember 2019, kell 09:29

2019-12-15T09:29:23Z

←Vanem redaktsioon		Redaktsioon: 15. detsember 2019, kell 11:29
23. rida:		23. rida:
	</tr>		</tr>
	</table>		</table>


			== <font color="red">Esmaspäeval ainult loeng kell 10:00</font> ==

			Esmaspäeval loengus kell 10 plaanime eksami sisu ja vaatame üle praktilised tööd.

			Kella 14-ne praktikum jääb ära: õppejõul samal ajal intervjuu err-le.

	== Exams ==		== Exams ==

Tammet – 9. detsember 2019, kell 12:06

2019-12-09T12:06:52Z

←Vanem redaktsioon		Redaktsioon: 9. detsember 2019, kell 14:06
23. rida:		23. rida:
	</tr>		</tr>
	</table>		</table>

			== Exams ==

			Kõik eksamid toimuvad esmaspäeviti kell 12:00 ja kell 15:00, konkreetne aeg vali ise:

			* 06.01.2019 12:00 ja 15:00 ruumis U02-102
			* 13.01.2019 12:00 ja 15:00 ruumis U06A-229
			* 20.01.2019 12:00 ja 15:00 ruumis U02-102

	== Time and place ==		== Time and place ==

Tammet: /* Practice work 6 */

2019-10-22T21:11:33Z

Practice work 6

←Vanem redaktsioon		Redaktsioon: 22. oktoober 2019, kell 23:11
219. rida:		219. rida:
	** https://gluebenchmark.com/		** https://gluebenchmark.com/
	** https://ai.googleblog.com/2019/01/natural-questions-new-corpus-and.html ja https://ai.google.com/research/NaturalQuestions		** https://ai.googleblog.com/2019/01/natural-questions-new-corpus-and.html ja https://ai.google.com/research/NaturalQuestions
	** A ~~dissertation~~ concerning answering winograd schema: https://repository.asu.edu/attachments/143311/content/Sharma_asu_0010N_14458.pdf and a paper by the author and others https://www.ijcai.org/Proceedings/15/Papers/190.pdf		** A master thesis concerning answering winograd schema: https://repository.asu.edu/attachments/143311/content/Sharma_asu_0010N_14458.pdf and a paper by the author and others https://www.ijcai.org/Proceedings/15/Papers/190.pdf
	** https://digital.lib.washington.edu/researchworks/bitstream/handle/1773/26336/Fader_washington_0250E_13164.pdf?sequence=1&isAllowed=y		** https://digital.lib.washington.edu/researchworks/bitstream/handle/1773/26336/Fader_washington_0250E_13164.pdf?sequence=1&isAllowed=y
	** https://www.aclweb.org/anthology/D18-1455/		** https://www.aclweb.org/anthology/D18-1455/

Tammet: /* Practice work 6 */

2019-10-22T21:10:56Z

Practice work 6

←Vanem redaktsioon		Redaktsioon: 22. oktoober 2019, kell 23:10
219. rida:		219. rida:
	** https://gluebenchmark.com/		** https://gluebenchmark.com/
	** https://ai.googleblog.com/2019/01/natural-questions-new-corpus-and.html ja https://ai.google.com/research/NaturalQuestions		** https://ai.googleblog.com/2019/01/natural-questions-new-corpus-and.html ja https://ai.google.com/research/NaturalQuestions
	** A dissertation concerning answering winograd schema: https://repository.asu.edu/attachments/143311/content/Sharma_asu_0010N_14458.pdf		** A dissertation concerning answering winograd schema: https://repository.asu.edu/attachments/143311/content/Sharma_asu_0010N_14458.pdf and a paper by the author and others https://www.ijcai.org/Proceedings/15/Papers/190.pdf
	** https://digital.lib.washington.edu/researchworks/bitstream/handle/1773/26336/Fader_washington_0250E_13164.pdf?sequence=1&isAllowed=y		** https://digital.lib.washington.edu/researchworks/bitstream/handle/1773/26336/Fader_washington_0250E_13164.pdf?sequence=1&isAllowed=y
	** https://www.aclweb.org/anthology/D18-1455/		** https://www.aclweb.org/anthology/D18-1455/

Tammet: /* Practice work 6 */

2019-10-22T21:01:52Z

Practice work 6

←Vanem redaktsioon		Redaktsioon: 22. oktoober 2019, kell 23:01
219. rida:		219. rida:
	** https://gluebenchmark.com/		** https://gluebenchmark.com/
	** https://ai.googleblog.com/2019/01/natural-questions-new-corpus-and.html ja https://ai.google.com/research/NaturalQuestions		** https://ai.googleblog.com/2019/01/natural-questions-new-corpus-and.html ja https://ai.google.com/research/NaturalQuestions
			** A dissertation concerning answering winograd schema: https://repository.asu.edu/attachments/143311/content/Sharma_asu_0010N_14458.pdf
	** https://digital.lib.washington.edu/researchworks/bitstream/handle/1773/26336/Fader_washington_0250E_13164.pdf?sequence=1&isAllowed=y		** https://digital.lib.washington.edu/researchworks/bitstream/handle/1773/26336/Fader_washington_0250E_13164.pdf?sequence=1&isAllowed=y
	** https://www.aclweb.org/anthology/D18-1455/		** https://www.aclweb.org/anthology/D18-1455/

Tammet: /* Practice work 6 */

2019-10-22T20:54:26Z

Practice work 6

←Vanem redaktsioon		Redaktsioon: 22. oktoober 2019, kell 22:54
188. rida:		188. rida:

	* Write a standalone implementation for selecting potentially useful rules/facts from a large set of axioms, using both ideas from SinE (see http://resources.mpi-inf.mpg.de/departments/rg1/conferences/deduction10/slides/krystof-hoder.pdf ) and word vectors (see https://link.springer.com/chapter/10.1007/978-3-030-29436-6_15 as a starting point)		* Write a standalone implementation for selecting potentially useful rules/facts from a large set of axioms, using both ideas from SinE (see http://resources.mpi-inf.mpg.de/departments/rg1/conferences/deduction10/slides/krystof-hoder.pdf ) and word vectors (see https://link.springer.com/chapter/10.1007/978-3-030-29436-6_15 as a starting point)

	It may be useful to google AMR parsing. The main direct link: https://amr.isi.edu/language.html and one actual parser https://github.com/mdtux89/amr-eager . See also https://paperswithcode.com/task/amr-parsing

	* Find out if there is a feasible way to programmatically build a large set of default properties for most popular noun categories in wordnet/wikipedia, a la "physical things typically cannot fly", "bird typically can fly", "penguins typically cannot fly", "dogs are typically hairy" etc. Start from looking at rules in conceptnet/nell/etc and continue looking at the beginnings of the wikipedia articles: is there hope of automatic parsing of texts to get such properties. Also, maybe they can be gleaned from existing word vectors?		* Find out if there is a feasible way to programmatically build a large set of default properties for most popular noun categories in wordnet/wikipedia, a la "physical things typically cannot fly", "bird typically can fly", "penguins typically cannot fly", "dogs are typically hairy" etc. Start from looking at rules in conceptnet/nell/etc and continue looking at the beginnings of the wikipedia articles: is there hope of automatic parsing of texts to get such properties. Also, maybe they can be gleaned from existing word vectors?
197. rida:		195. rida:

	* Investigating available frameworks/tools suitable to be used for converting natural language (English) to a logical formalism, assuming it is ok to use confidence of statements/rules, default logic, word vector based similarities and other such extensions as necessary. After selection, using available tools to build a prototype converter.		* Investigating available frameworks/tools suitable to be used for converting natural language (English) to a logical formalism, assuming it is ok to use confidence of statements/rules, default logic, word vector based similarities and other such extensions as necessary. After selection, using available tools to build a prototype converter.

			It may be useful to google AMR parsing. The main direct link: https://amr.isi.edu/language.html and one actual parser https://github.com/mdtux89/amr-eager . See also https://paperswithcode.com/task/amr-parsing
			* Have a brief look at
			** https://nlp.stanford.edu/software/sempre/
			** https://www.aclweb.org/anthology/P15-1129.pdf
			** https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/44858.pdf
			** https://yoavartzi.com/sp14/slides/mooney.sp14.pdf

	* Investigating and converting http://conceptnet5.media.mit.edu/ to a logical formalism, using all the non-standard representation mechanisms (again, onfidence of statements/rules, default logic, word vector based similarities and other such extensions as necessary). A TPTP dump of conceptnet is available. Ideally integrating the ruleset with https://wordnet.princeton.edu/ . A TPTP dump of a taxonomy of wordnet is available. It may be a good idea to select only rules for the "most common" words: the "commonness" table is available.		* Investigating and converting http://conceptnet5.media.mit.edu/ to a logical formalism, using all the non-standard representation mechanisms (again, onfidence of statements/rules, default logic, word vector based similarities and other such extensions as necessary). A TPTP dump of conceptnet is available. Ideally integrating the ruleset with https://wordnet.princeton.edu/ . A TPTP dump of a taxonomy of wordnet is available. It may be a good idea to select only rules for the "most common" words: the "commonness" table is available.

Tammet: /* Practice work 6 */

2019-10-22T20:35:11Z

Practice work 6

←Vanem redaktsioon		Redaktsioon: 22. oktoober 2019, kell 22:35
188. rida:		188. rida:

	* Write a standalone implementation for selecting potentially useful rules/facts from a large set of axioms, using both ideas from SinE (see http://resources.mpi-inf.mpg.de/departments/rg1/conferences/deduction10/slides/krystof-hoder.pdf ) and word vectors (see https://link.springer.com/chapter/10.1007/978-3-030-29436-6_15 as a starting point)		* Write a standalone implementation for selecting potentially useful rules/facts from a large set of axioms, using both ideas from SinE (see http://resources.mpi-inf.mpg.de/departments/rg1/conferences/deduction10/slides/krystof-hoder.pdf ) and word vectors (see https://link.springer.com/chapter/10.1007/978-3-030-29436-6_15 as a starting point)

			It may be useful to google AMR parsing. The main direct link: https://amr.isi.edu/language.html and one actual parser https://github.com/mdtux89/amr-eager . See also https://paperswithcode.com/task/amr-parsing

	* Find out if there is a feasible way to programmatically build a large set of default properties for most popular noun categories in wordnet/wikipedia, a la "physical things typically cannot fly", "bird typically can fly", "penguins typically cannot fly", "dogs are typically hairy" etc. Start from looking at rules in conceptnet/nell/etc and continue looking at the beginnings of the wikipedia articles: is there hope of automatic parsing of texts to get such properties. Also, maybe they can be gleaned from existing word vectors?		* Find out if there is a feasible way to programmatically build a large set of default properties for most popular noun categories in wordnet/wikipedia, a la "physical things typically cannot fly", "bird typically can fly", "penguins typically cannot fly", "dogs are typically hairy" etc. Start from looking at rules in conceptnet/nell/etc and continue looking at the beginnings of the wikipedia articles: is there hope of automatic parsing of texts to get such properties. Also, maybe they can be gleaned from existing word vectors?