博碩士論文 111526012 完整後設資料紀錄

DC 欄位 語言
DC.contributor資訊工程學系zh_TW
DC.creator張翔威zh_TW
DC.creatorHsiang-Wei Changen_US
dc.date.accessioned2025-1-9T07:39:07Z
dc.date.available2025-1-9T07:39:07Z
dc.date.issued2025
dc.identifier.urihttp://ir.lib.ncu.edu.tw:444/thesis/view_etd.asp?URN=111526012
dc.contributor.department資訊工程學系zh_TW
DC.description國立中央大學zh_TW
DC.descriptionNational Central Universityen_US
dc.description.abstract本研究旨在探索和實現基於GPT-2和近端策略優化(PPO)強化學習的 符號音樂生成方法。音樂生成在人工智慧與機器學習領域中一直是重要 的研究課題之一,目的是通過算法自動生成具有藝術價值和情感表達的 音樂作品。本研究通過探討GPT-2在生成旋律、伴奏及和弦方面的應用潛 力,並引入PPO強化學習算法,提升生成音樂的質量和多樣性。研究結果 表明,結合GPT-2和PPO的符號音樂生成方法能夠有效提升音樂的創造性 和多樣性,為音樂創作和應用帶來新的可能性。本研究不僅在理論上具有 重要意義,還在實際應用中具有廣泛的前景。zh_TW
dc.description.abstractThis study aims to explore and implement a symbolic music generation method based on GPT-2 and Proximal Policy Optimization (PPO) reinforcement learning. Music gen eration has long been an important research topic in the fields of artificial intelligence and machine learning, aiming to automatically generate music works with artistic value and emotional expression through algorithms. This research explores the application potential of GPT-2 in generating melodies, accompaniments, and chords, and introduces the PPO reinforcement learning algorithm to improve the quality and diversity of gen erated music. The results show that the symbolic music generation method combining GPT-2 and PPO can effectively enhance the creativity and diversity of music, bringing new possibilities to music creation and application. This research is of great theoretical significance and has broad prospects for practical application.en_US
DC.subject音樂生成zh_TW
DC.subject近端策略優化zh_TW
DC.subject強化學習zh_TW
DC.subject符號音樂生成zh_TW
DC.subjectGPT-2en_US
DC.title基於GPT-2和近端策略優化的符號音樂生成zh_TW
dc.language.isozh-TWzh-TW
DC.titleSymbolic Music Generation Using GPT-2 and Proximal Policy Optimizationen_US
DC.type博碩士論文zh_TW
DC.typethesisen_US
DC.publisherNational Central Universityen_US

若有論文相關問題,請聯絡國立中央大學圖書館推廣服務組 TEL:(03)422-7151轉57407,或E-mail聯絡  - 隱私權政策聲明