東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

FindBook

Google Book

Amazon

博客來

Unsupervised Generative Chord Representation Learning and its Effect on Novelty-Creativity and Fidelity-Standards.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Unsupervised Generative Chord Representation Learning and its Effect on Novelty-Creativity and Fidelity-Standards./
作者:	Valladares, Agustin Arnoldo Macaya.
面頁冊數:	1 online resource (75 pages)
附註:	Source: Masters Abstracts International, Volume: 84-11.
Contained By:	Masters Abstracts International84-11.
標題:	Piano. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30456061click for full text (PQDT)
ISBN:	9798379487706

Unsupervised Generative Chord Representation Learning and its Effect on Novelty-Creativity and Fidelity-Standards.
Valladares, Agustin Arnoldo Macaya.

Unsupervised Generative Chord Representation Learning and its Effect on Novelty-Creativity and Fidelity-Standards. - 1 online resource (75 pages)

Source: Masters Abstracts International, Volume: 84-11.

Thesis (M.Sc.)--Pontificia Universidad Catolica de Chile (Chile), 2023.

Includes bibliographical references

Generative models in deep learning have experienced great development in art generation. Even though image-based art generation has had a big success, music still needs to catch up compared to its counterpart. Extreme focus on improving the generated outputs has neglected the importance of understanding what generative models learn and understand about music. This investigation aims to understand how latent space characteristics in generative models affect the generation of novel musical examples and their relationship to musical concept learning. Unsupervised VAEs with different latent space characteristics were trained to generate chords. Reconstruction and generation capabilities were analyzed. A set of probing networks were trained to determine the representations learned by the unsupervised models. Particular focus was drawn to identify which musical concepts were learned in the latent space. Analysis shows that a bigger latent space will favor, with limitations, novelty-creativity at the expense of fidelity-standards, which get worse but also to a limit. Other findings show that smaller latent spaces do not allow for good dataset reconstruction but still follow good fidelity-standards at generation time at the expense of lower novelty-creativity. Finally, results show that bigger latent spaces are required for learning complex musical concepts.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2023

Mode of access: World Wide Web

ISBN: 9798379487706Subjects--Topical Terms:

526062
Piano.
Index Terms--Genre/Form:

542853
Electronic books.

Unsupervised Generative Chord Representation Learning and its Effect on Novelty-Creativity and Fidelity-Standards.
LDR:04241nmm a2200373K 4500 001 2361381
005 20231019120610.5
006 m o d
007 cr mn ---uuuuu
008 241011s2023 xx obm 000 0 eng d
020 $a 9798379487706
035 $a (MiAaPQ)AAI30456061
035 $a (MiAaPQ)PontificiaChile_1153466522
035 $a AAI30456061
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Valladares, Agustin Arnoldo Macaya. $3 3702052
245 1 0 $a Unsupervised Generative Chord Representation Learning and its Effect on Novelty-Creativity and Fidelity-Standards.
264 0 $c 2023
300 $a 1 online resource (75 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Masters Abstracts International, Volume: 84-11.
500 $a Advisor: Cadiz, Rodrigo;Parra, Denis.
502 $a Thesis (M.Sc.)--Pontificia Universidad Catolica de Chile (Chile), 2023.
504 $a Includes bibliographical references
520 $a Generative models in deep learning have experienced great development in art generation. Even though image-based art generation has had a big success, music still needs to catch up compared to its counterpart. Extreme focus on improving the generated outputs has neglected the importance of understanding what generative models learn and understand about music. This investigation aims to understand how latent space characteristics in generative models affect the generation of novel musical examples and their relationship to musical concept learning. Unsupervised VAEs with different latent space characteristics were trained to generate chords. Reconstruction and generation capabilities were analyzed. A set of probing networks were trained to determine the representations learned by the unsupervised models. Particular focus was drawn to identify which musical concepts were learned in the latent space. Analysis shows that a bigger latent space will favor, with limitations, novelty-creativity at the expense of fidelity-standards, which get worse but also to a limit. Other findings show that smaller latent spaces do not allow for good dataset reconstruction but still follow good fidelity-standards at generation time at the expense of lower novelty-creativity. Finally, results show that bigger latent spaces are required for learning complex musical concepts.
520 $a Los modelos generativos de aprendizaje profundo han experimentado un gran desarrollo en la generacion de arte. Aunque la generacion de arte basada en imagenes ha tenido un gran exito, la musica todavia tiene mucho por delante para alcanzar a su contraparte. Especial enfoque en mejorar los resultados generados ha descuidado la importancia de comprender como los modelos generativos aprenden y entienden la musica. Esta investigacion tiene como objetivo comprender como las caracteristicas del espacio latente en los modelos generativos afectan la generacion de ejemplos musicales novedosos y su relacion con el aprendizaje de conceptos musicales. Se entrenaron VAEs no-supervisadas con espacios latentes de diferentes caracteristicas para generar acordes. Se analizaron las capacidades de reconstruccion y generacion. Se entreno un conjunto de redes de sondeo para determinar las representaciones aprendidas por los modelos no-supervisados. El analisis muestra que espacios latentes grandes favorecen, con limitaciones, la novedad-creatividad en perjuicio de la fidelidad-estandares, que empeoran pero con un limite. Otros hallazgos muestran espacios latentes pequenos no permiten una buena reconstruccion de los datos de entrenamiento, pero aun asi son suficientes para mantener la fidelidad-estandares en el momento de la generacion, a expensas de una menor novedad-creatividad. Finalmente, los resultados muestran que se requieren espacios latentes grandes para un mayor aprendizaje de conceptos musicales complejos.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2023
538 $a Mode of access: World Wide Web
650 4 $a Piano. $3 526062
650 4 $a Deep learning. $3 3554982
650 4 $a Success. $3 518195
650 4 $a Neural networks. $3 677449
650 4 $a Creativity. $3 605632
650 4 $a Chords (Music). $3 3697399
650 4 $a Sheet music. $3 3702053
650 4 $a Sound. $3 542298
650 4 $a Acoustics. $3 879105
650 4 $a Fine arts. $3 2122690
650 4 $a Music. $3 516178
655 7 $a Electronic books. $2 lcsh $3 542853
690 $a 0800
690 $a 0986
690 $a 0357
690 $a 0413
710 2 $a ProQuest Information and Learning Co. $3 783688
710 2 $a Pontificia Universidad Catolica de Chile (Chile). $3 3550823
773 0 $t Masters Abstracts International $g 84-11.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30456061 $z click for full text (PQDT)