المؤلفون: Karemaker, Valentijn, Yu, Chia-Fu

مصطلحات موضوعية: Mathematics - Number Theory, Mathematics - Algebraic Geometry, 14K10 (14K15, 11G10, 51A50)

الوصف: Let $\mathcal{A}_g$ be the moduli space over $\overline{\mathbb{F}}_p$ of $g$-dimensional principally polarised abelian varieties, where $p$ is a prime. We show that if $g$ is even and $p\geq 5$, then every geometric generic member in the maximal supersingular Ekedahl-Oort stratum in $\mathcal{A}_g$ has automorphism group $\{ \pm 1\}$. This confirms Oort's conjecture in the case of $p\geq 5$ and even $g$.
Comment: 31 pages

الوصول الحر: http://arxiv.org/abs/2406.19748Test

المؤلفون: Hung, Tzu-Yun, Wu, Jui-Te, Kuo, Yu-Chia, Hsiao, Yo-Wei, Lin, Ting-Wei, Su, Li

مصطلحات موضوعية: Computer Science - Sound, Computer Science - Multimedia, Electrical Engineering and Systems Science - Audio and Speech Processing

الوصف: Expressive music synthesis (EMS) for violin performance is a challenging task due to the disagreement among music performers in the interpretation of expressive musical terms (EMTs), scarcity of labeled recordings, and limited generalization ability of the synthesis model. These challenges create trade-offs between model effectiveness, diversity of generated results, and controllability of the synthesis system, making it essential to conduct a comparative study on EMS model design. This paper explores two violin EMS approaches. The end-to-end approach is a modification of a state-of-the-art text-to-speech generator. The parameter-controlled approach is based on a simple parameter sampling process that can render note lengths and other parameters compatible with MIDI-DDSP. We study these two approaches (in total, three model variants) through objective and subjective experiments and discuss several key issues of EMS based on the results.
Comment: 15 pages, 2 figures, 3 tables

الوصول الحر: http://arxiv.org/abs/2406.18089Test

المؤلفون: Huang, Yu-Fen, Moran, Nikki, Coleman, Simon, Kelly, Jon, Wei, Shun-Hwa, Chen, Po-Yin, Huang, Yun-Hsin, Chen, Tsung-Ping, Kuo, Yu-Chia, Wei, Yu-Chi, Li, Chih-Hsuan, Huang, Da-Yu, Kao, Hsuan-Kai, Lin, Ting-Wei, Su, Li

مصطلحات موضوعية: Computer Science - Sound, Computer Science - Artificial Intelligence, Electrical Engineering and Systems Science - Audio and Speech Processing

الوصف: In cross-modal music processing, translation between visual, auditory, and semantic content opens up new possibilities as well as challenges. The construction of such a transformative scheme depends upon a benchmark corpus with a comprehensive data infrastructure. In particular, the assembly of a large-scale cross-modal dataset presents major challenges. In this paper, we present the MOSA (Music mOtion with Semantic Annotation) dataset, which contains high quality 3-D motion capture data, aligned audio recordings, and note-by-note semantic annotations of pitch, beat, phrase, dynamic, articulation, and harmony for 742 professional music performances by 23 professional musicians, comprising more than 30 hours and 570 K notes of data. To our knowledge, this is the largest cross-modal music dataset with note-level annotations to date. To demonstrate the usage of the MOSA dataset, we present several innovative cross-modal music information retrieval (MIR) and musical content generation tasks, including the detection of beats, downbeats, phrase, and expressive contents from audio, video and motion data, and the generation of musicians' body motion from given music audio. The dataset and codes are available alongside this publication (https://github.com/yufenhuang/MOSA-Music-mOtion-and-Semantic-Annotation-datasetTest).
Comment: IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2024. 14 pages, 7 figures. Dataset is available on: https://github.com/yufenhuang/MOSA-Music-mOtion-and-Semantic-Annotation-dataset/tree/mainTest and https://zenodo.org/records/11393449Test

الوصول الحر: http://arxiv.org/abs/2406.06375Test

المؤلفون: Tsai, Yu-Lin, Li, Yizhe, Chen, Zekai, Chen, Po-Yu, Yu, Chia-Mu, Ren, Xuebin, Buet-Golfouse, Francois

مصطلحات موضوعية: Computer Science - Computer Vision and Pattern Recognition, Computer Science - Artificial Intelligence, Computer Science - Cryptography and Security

الوصف: The integration of Differential Privacy (DP) with diffusion models (DMs) presents a promising yet challenging frontier, particularly due to the substantial memorization capabilities of DMs that pose significant privacy risks. Differential privacy offers a rigorous framework for safeguarding individual data points during model training, with Differential Privacy Stochastic Gradient Descent (DP-SGD) being a prominent implementation. Diffusion method decomposes image generation into iterative steps, theoretically aligning well with DP's incremental noise addition. Despite the natural fit, the unique architecture of DMs necessitates tailored approaches to effectively balance privacy-utility trade-off. Recent developments in this field have highlighted the potential for generating high-quality synthetic data by pre-training on public data (i.e., ImageNet) and fine-tuning on private data, however, there is a pronounced gap in research on optimizing the trade-offs involved in DP settings, particularly concerning parameter efficiency and model scalability. Our work addresses this by proposing a parameter-efficient fine-tuning strategy optimized for private diffusion models, which minimizes the number of trainable parameters to enhance the privacy-utility trade-off. We empirically demonstrate that our method achieves state-of-the-art performance in DP synthesis, significantly surpassing previous benchmarks on widely studied datasets (e.g., with only 0.47M trainable parameters, achieving a more than 35% improvement over the previous state-of-the-art with a small privacy budget on the CelebA-64 dataset). Anonymous codes available at https://anonymous.4open.science/r/DP-LORA-F02FTest.
Comment: 16 pages, 5 figures, 11 tables

الوصول الحر: http://arxiv.org/abs/2406.01355Test

المؤلفون: Hsu, Chia-Yi, Tsai, Yu-Lin, Lin, Chih-Hsun, Chen, Pin-Yu, Yu, Chia-Mu, Huang, Chun-Ying

مصطلحات موضوعية: Computer Science - Machine Learning

الوصف: While large language models (LLMs) such as Llama-2 or GPT-4 have shown impressive zero-shot performance, fine-tuning is still necessary to enhance their performance for customized datasets, domain-specific tasks, or other private needs. However, fine-tuning all parameters of LLMs requires significant hardware resources, which can be impractical for typical users. Therefore, parameter-efficient fine-tuning such as LoRA have emerged, allowing users to fine-tune LLMs without the need for considerable computing resources, with little performance degradation compared to fine-tuning all parameters. Unfortunately, recent studies indicate that fine-tuning can increase the risk to the safety of LLMs, even when data does not contain malicious content. To address this challenge, we propose Safe LoRA, a simple one-liner patch to the original LoRA implementation by introducing the projection of LoRA weights from selected layers to the safety-aligned subspace, effectively reducing the safety risks in LLM fine-tuning while maintaining utility. It is worth noting that Safe LoRA is a training-free and data-free approach, as it only requires the knowledge of the weights from the base and aligned LLMs. Our extensive experiments demonstrate that when fine-tuning on purely malicious data, Safe LoRA retains similar safety performance as the original aligned model. Moreover, when the fine-tuning dataset contains a mixture of both benign and malicious data, Safe LoRA mitigates the negative effect made by malicious data while preserving performance on downstream tasks.

الوصول الحر: http://arxiv.org/abs/2405.16833Test

المؤلفون: Hung, Fan-Yun, Yu, Chia-Fu

مصطلحات موضوعية: Mathematical Physics, Numerical and Computational Mathematics, Pure Mathematics, Mathematical Sciences, Class numbers, Multinorm-one tori, General Mathematics, Pure mathematics

وصف الملف: application/pdf

الوصول الحر: https://escholarship.org/uc/item/9n16q2jvTest

المؤلفون: Marjit, Shyam, Singh, Harshit, Mathur, Nityanand, Paul, Sayak, Yu, Chia-Mu, Chen, Pin-Yu

مصطلحات موضوعية: Computer Science - Computer Vision and Pattern Recognition

الوصف: In the realm of subject-driven text-to-image (T2I) generative models, recent developments like DreamBooth and BLIP-Diffusion have led to impressive results yet encounter limitations due to their intensive fine-tuning demands and substantial parameter requirements. While the low-rank adaptation (LoRA) module within DreamBooth offers a reduction in trainable parameters, it introduces a pronounced sensitivity to hyperparameters, leading to a compromise between parameter efficiency and the quality of T2I personalized image synthesis. Addressing these constraints, we introduce \textbf{\textit{DiffuseKronA}}, a novel Kronecker product-based adaptation module that not only significantly reduces the parameter count by 35\% and 99.947\% compared to LoRA-DreamBooth and the original DreamBooth, respectively, but also enhances the quality of image synthesis. Crucially, \textit{DiffuseKronA} mitigates the issue of hyperparameter sensitivity, delivering consistent high-quality generations across a wide range of hyperparameters, thereby diminishing the necessity for extensive fine-tuning. Furthermore, a more controllable decomposition makes \textit{DiffuseKronA} more interpretable and even can achieve up to a 50\% reduction with results comparable to LoRA-Dreambooth. Evaluated against diverse and complex input images and text prompts, \textit{DiffuseKronA} consistently outperforms existing models, producing diverse images of higher quality with improved fidelity and a more accurate color distribution of objects, all the while upholding exceptional parameter efficiency, thus presenting a substantial advancement in the field of T2I generative modeling. Our project page, consisting of links to the code, and pre-trained checkpoints, is available at https://diffusekrona.github.ioTest/.
Comment: Project Page: https://diffusekrona.github.ioTest/

الوصول الحر: http://arxiv.org/abs/2402.17412Test

المؤلفون: Karemaker, Valentijn, Tamagawa, Akio, Yu, Chia-Fu

مصطلحات موضوعية: Mathematics - Number Theory, Mathematics - Algebraic Geometry, 16H20 (14K12 11E39 11G10)

الوصف: We prove uniqueness of a decomposition of $1$ into indecomposable Hermitian idempotents in an order of a finite-dimensional $\mathbb{Q}$-algebra with positive involution, by generalising a result of Eichler on unique decomposition of lattices. We use this result to prove that polarised abelian varieties over any field admit a unique decomposition into indecomposable polarised abelian subvarieties, a result previously shown by Debarre and Serre with different methods and over algebraically closed fields. We prove that an analogous uniqueness result holds true for arbitrary polarised integral Hodge structures, and derive a consequence for their automorphism groups.
Comment: 12 pages

الوصول الحر: http://arxiv.org/abs/2402.09323Test

المؤلفون: Liang, Pei-Xin, Oki, Yasuhiro, Yu, Chia-Fu

مصطلحات موضوعية: Mathematics - Number Theory

الوصف: In this paper we investigate the Tate--Shafarevich group Sha^1(k, T) of a multinorm-one torus $T$ over a global field $k$. We establish a few functorial maps among cohomology groups and explore their relations. Using these properties and relations we obtain a few basic structural results for Sha^1(k, T) and extend a few results of Bayer-Fluckiger--Lee--Parimala [Adv. in Math., 2019] to some more general multinorm-one tori. We also give a uniform proof of a result of Demarche--Wei for a criterion of the vanishing of Sha^1(k, T), and of the main result of Pollio [Pure App. Math. Q., 2014] for the case where the \'etale $k$-algebra in question is a product of two abelian extensions. Moreover, we improve the explicit description of Sha^1(k, T) in Lee [J. Pure Appl. Alg., 2022] by removing an intersection condition.
Comment: 26 pages, comments welcome

الوصول الحر: http://arxiv.org/abs/2310.20133Test

المؤلفون: Su, Kate Y. L., Kennedy, Grant M., Rieke, George H., Hughes, A. Meredith, Lin, Yu-Chia, Kittling, Jamar, Jackson, Alan P., Anche, Ramya M., Liu, Hauyu Baobab

مصطلحات موضوعية: Astrophysics - Earth and Planetary Astrophysics, Astrophysics - Solar and Stellar Astrophysics

الوصف: RZ Piscium (RZ Psc) is well-known in the variable star field because of its numerous, irregular optical dips in the past five decades, but the nature of the system is heavily debated in the literature. We present multiyear infrared monitoring data from Spitzer and WISE to track the activities of the inner debris production, revealing stochastic infrared variability as short as weekly timescales that is consistent with destroying a 90-km-size asteroid every year. ALMA 1.3 mm data combined with spectral energy distribution modeling show that the disk is compact ($\sim$0.1--13 au radially) and lacks cold gas. The disk is found to be highly inclined and has a significant vertical scale height. These observations confirm that RZ Psc hosts a close to edge-on, highly perturbed debris disk possibly due to migration of recently formed giant planets which might be triggered by the low-mass companion RZ Psc B if the planets formed well beyond the snowlines.
Comment: 16 pages, 5 figures, accepted for publication in ApJ

الوصول الحر: http://arxiv.org/abs/2310.12336Test