检索结果-内蒙古大学图书馆

Towards generalizing deep-audio fake detection networks

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Gasenzer, Konstantin Wolter, Moritz High Performance Computing and Analytics Lab University of Bonn Germany

Today's generative neural networks allow the creation of high-quality synthetic speech at scale. While we welcome the creative use of this new technology, we must also recognize the risks. As synthetic speech is abused for monetary and identity theft, we require a broad set of deepfake identification tools. Furthermore, previous work reported a limited ability of deep classifiers to generalize to unseen audio generators. We study the frequency domain fingerprints of current audio generators. Building on top of the discovered frequency footprints, we train excellent lightweight detectors that generalize. We report improved results on the WaveFake dataset and an extended version. To account for the rapid progress in the field, we extend the WaveFake dataset by additionally considering samples drawn from the novel Avocodo and BigVGAN networks. For illustration purposes, the supplementary material contains audio samples of generator artifacts. © 2023, CC BY-NC-SA.

关键词： Frequency domain analysis

ptwt - the PyTorch wavelet toolbox

学校读者我要写书评

暂无评论

The Journal of Machine Learning Research

The Journal of Machine Learning Research 2024年第1期25卷 3892-3898页

作者： Moritz Wolter Felix Blanke Jochen Garcke Charles Tapley Hoyt High-Performance Computing and Analytics Lab University of Bonn Germany Fraunhofer Institute for Algorithms and Scientific Computing Sankt Augustin Germany Institute for Numerical Simulation University of Bonn and Fraunhofer Institute for Algorithms and Scientific Computing Sankt Augustin Germany Northeastern University Boston

The fast wavelet transform is an important workhorse in signal processing. Wavelets are local in the spatial- or temporal- and the frequency-domain. This property enables frequency domain analysis while preserving some spatiotemporal information. Until recently, wavelets rarely appeared in the machine learning literature. We provide the PyTorch Wavelet Toolbox to make wavelet methods more accessible to the deep learning community. Our PyTorch Wavelet Toolbox is well documented. A pip package is installable with pip install ptwt.

关键词： PyTorch wavelet wavelet-packets wavelet-analysis wavelet-transform

Correction to: Inductive linearization for binary quadratic programs with linear constraints

学校读者我要写书评

暂无评论

4OR 2021年第4期19卷 638-639页

作者： Mallach, Sven High Performance Computing and Analytics Lab University of Bonn Bonn Germany

Twisted mass ensemble generation on GPU machines 39

学校读者我要写书评

暂无评论

Twisted mass ensemble generation on GPU machines

39th International Symposium on Lattice Field Theory, LATTICE 2022

作者： Kostrzewa, Bartosz Bacchio, Simone Finkenrath, Jacob Garofalo, Marco Pittler, Ferenc Romiti, Simone Urbach, Carsten High-Performance Computing and Analytics Lab Rheinische Friedrich-Wilhelms-Universität Bonn Friedrich-Hirzebruch-Allee 8 Bonn53115 Germany Bethe Center for Theoretical Physics Rheinische Friedrich-Wilhelms-Universität Bonn Nussallee 14-16 Bonn53115 Germany The Cyprus Institute CASTORC 20 Konstantinou Kavafi Street Nicosia2121 Cyprus

We present how we ported the Hybrid Monte Carlo implementation in the tmLQCD software suite to GPUs through offloading its most expensive parts to the QUDA library. We discuss our motivations and some of the technical challenges that we encountered as we added the required functionality to both tmLQCD and QUDA. We further present some performance details, focussing in particular on the usage of QUDA's multigrid solver for poorly conditioned light quark monomials as well as the multi-shift solver for the non-degenerate strange and charm sector in Nf = 2 + 1 + 1 simulations using twisted mass clover fermions, comparing the efficiency of state-of-the-art simulations on CPU and GPU machines. We also take a look at the performance-portability question through preliminary tests of our HMC on a machine based on AMD's MI250 GPU, finding good performance after a very minor additional porting effort. Finally, we conclude that we should be able to achieve GPU utilisation factors acceptable for the current generation of (pre-)exascale supercomputers with subtantial efficiency improvements and real time speedups compared to just running on CPUs. At the same time, we find that future challenges will require different approaches and, most importantly, a very significant investment of personnel for software development. © Copyright owned by the author(s) under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0).

关键词： Graphics processing unit

Inclusive Hadronic Decay Rate of the τ Lepton from Lattice QCD: The u¯s Flavor Channel and the Cabibbo Angle

学校读者我要写书评

暂无评论

Physical Review Letters 2024年第26期132卷 261901-261901页

作者： Constantia Alexandrou Simone Bacchio Alessandro De Santis Antonio Evangelista Jacob Finkenrath Roberto Frezzotti Giuseppe Gagliardi Marco Garofalo Bartosz Kostrzewa Vittorio Lubicz Simone Romiti Francesco Sanfilippo Silvano Simula Nazario Tantalo Carsten Urbach Urs Wenger Department of Physics Computation-based Science and Technology Research Center Dipartimento di Fisica and Gaustrae 20 42119 Wuppertal Germany Istituto Nazionale di Fisica Nucleare HISKP (Theory) High Performance Computing and Analytics Lab Dipartimento di Matematica e Fisica Institute for Theoretical Physics Albert Einstein Center for Fundamental Physics

We present a lattice determination of the inclusive decay rate of the process τ↦Xusντ in which the τ lepton decays into a generic hadronic state Xus with u¯s flavor quantum numbers. Our results have been obtained in nf=2+1+1 isosymmetric QCD with full nonperturbative accuracy, without any operator product expansion approximation and, except for the presently missing long-distance isospin-breaking corrections, include a solid estimate of all sources of theoretical uncertainties. This has been possible by using the Hansen-Lupo-Tantalo method [M. Hansen et al., Phys. Rev. D 99, 094508 (2019)] that we have already successfully applied [A. Evangelista et al., Phys. Rev. D 108, 074513 (2023)] to compute the inclusive decay rate of the process τ↦Xudντ in the u¯d flavor channel. By combining our first-principles theoretical results with the presently available experimental data, we extract the Cabibbo-Kobayashi-Maskawa matrix element |Vus|, the Cabibbo angle, with a 0.9% accuracy, dominated by the experimental error.

关键词： Cabibbo–Kobayashi–Maskawa matrix Hadronic decays Lattice field theory Lattice gauge theory Lepton-hadron interactions Non-Abelian gauge theories Nonperturbative effects in field theory Phenomenology Quantum chromodynamics Strong interaction Total cross sections

Fréchet Wavelet Distance: A Domain-Agnostic Metric for Image Generation

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Veeramacheneni, Lokesh Wolter, Moritz Kuehne, Hildegard Gall, Juergen High-Performance Computing and Analytics Lab University of Bonn Bonn53115 Germany Multimodal Learning Group University of Bonn Bonn53115 Germany Computer Vision Group University of Bonn Bonn53115 Germany

Modern metrics for generative learning like Fréchet Inception Distance (FID) demonstrate impressive performance. However, they suffer from various shortcomings, like a bias towards specific generators and datasets. To address this problem, we propose the Fréchet Wavelet Distance (FWD) as a domain-agnostic metric based on Wavelet Packet Transform (Wp). FWD provides a sight across a broad spectrum of frequencies in images with a high resolution, along with preserving both spatial and textural aspects. Specifically, we use Wp to project generated and dataset images to packet coefficient space. Further, we compute Fréchet distance with the resultant coefficients to evaluate the quality of a generator. This metric is general-purpose and dataset-domain agnostic, as it does not rely on any pre-trained network while being more interpretable because of frequency band transparency. We conclude with an extensive evaluation of a wide variety of generators across various datasets that the proposed FWD is able to generalize and improve robustness to domain shift and various corruptions compared to other metrics. © 2023, CC BY.

关键词： Quality control

Tucano: Advancing Neural Text Generation for Portuguese

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Corrêa, Nicholas Kluge Sen, Aniket Falk, Sophia Fatimah, Shiza The Center for Science and Thought the University of Bonn NRW Bonn Germany The High Performance Computing and Analytics Lab the Helmholtz-Institut für Strahlen- und Kernphysik the University of Bonn Germany The Bonn Sustainable AI Lab Institute for Science and Ethics University of Bonn Germany The Institute for Computer Science the University of Bonn Germany

Significant advances have been made in natural language processing in recent years. However, our current deep learning approach to language modeling requires substantial resources in terms of data and computation. One of the side effects of this data-hungry paradigm is the current schism between languages, separating those considered high-resource, where most of the development happens and resources are available, and the low-resource ones, which struggle to attain the same level of performance and autonomy. This study aims to introduce a new set of resources to stimulate the future development of neural text generation in Portuguese. In this work, we document the development of GigaVerbo, a concatenation of deduplicated Portuguese text corpora amounting to 200 billion tokens. Via this corpus, we trained a series of decoder-transformers named Tucano. Our models perform equal or superior to other Portuguese and multilingual language models of similar size in several Portuguese benchmarks. The evaluation of our models also reveals that model performance on many currently available benchmarks used by the Portuguese NLP community has little to no correlation with the scaling of token ingestion during training, highlighting the limitations of such evaluations when it comes to the assessment of Portuguese generative language models. All derivatives of our study are openly released on GitHub1 and Hugging Face2 © 2024, CC BY-NC-SA.

关键词： Modeling languages

Canonical convolutional neural networks

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Veeramacheneni, Lokesh Wolter, Moritz Klein, Reinhard Garcke, Jochen Department of Computer Science Hochschule Bonn-Rhein-Sieg Fraunhofer Center for Machine Learning SCAI Germany High Performance Computing and Analytics Lab University of Bonn Fraunhofer Center for Machine Learning SCAI Germany Department of Computer Science University of Bonn Germany Fraunhofer Center for Machine Learning SCAI Institute for Numerical Simulation University of Bonn Germany

We introduce canonical weight normalization for convolutional neural networks. Inspired by the canonical tensor decomposition, we express the weight tensors in so-called canonical networks as scaled sums of outer vector products. In particular, we train network weights in the decomposed form, where scale weights are optimized separately for each mode. Additionally, similarly to weight normalization, we include a global scaling parameter. We study the initialization of the canonical form by running the power method and by drawing randomly from Gaussian or uniform distributions. Our results indicate that we can replace the power method with cheaper initializations drawn from standard distributions. The canonical re-parametrization leads to competitive normalization performance on the MNIST, CIFAR10, and SVHN data sets. Moreover, the formulation simplifies network compression. Once training has converged, the canonical form allows convenient model-compression by truncating the parameter sums. © 2022, CC BY-NC-SA.

关键词： Convolutional neural networks

Twisted mass ensemble generation on GPU machines

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Kostrzewa, Bartosz Bacchio, Simone Finkenrath, Jacob Garofalo, Marco Pittler, Ferenc Romitiband, Simone Urbach, Carsten High-Performance Computing and Analytics Lab Rheinische Friedrich-Wilhelms-Universität Bonn Friedrich-Hirzebruch-Allee 8 Bonn53115 Germany Bethe Center for Theoretical Physics Rheinische Friedrich-Wilhelms-Universität Bonn Nussallee 14-16 Bonn53115 Germany The Cyprus Institute CASTORC 20 Konstantinou Kavafi Street Nicosia2121 Cyprus

We present how we ported the Hybrid Monte Carlo implementation in the tmLQCD software suite to GPUs through offloading its most expensive parts to the QUDA library. We discuss our motivations and some of the technical challenges that we encountered as we added the required functionality to both tmLQCD and QUDA. We further present some performance details, focussing in particular on the usage of QUDA’s multigrid solver for poorly conditioned light quark monomials as well as the multi-shift solver for the non-degenerate strange and charm sector in Nf = 2 + 1 + 1 simulations using twisted mass clover fermions, comparing the efficiency of state-of-the-art simulations on CPU and GPU machines. We also take a look at the performance-portability question through preliminary tests of our HMC on a machine based on AMD’s MI250 GPU, finding good performance after a very minor additional porting effort. Finally, we conclude that we should be able to achieve GPU utilisation factors acceptable for the current generation of (pre-)exascale supercomputers with subtantial efficiency improvements and real time speedups compared to just running on CPUs. At the same time, we find that future challenges will require different approaches and, most importantly, a very significant investment of personnel for software development. © 2022, CC BY-NC-ND.

关键词： Graphics processing unit