检索结果-内蒙古大学图书馆

EXPRESSIVITY AND SPEECH SYNTHESIS

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Triantafyllopoulos, Andreas Schuller, Björn W. Technical University of Munich MRI Munich Germany GLAM – Group on Language Audio & Music Imperial College London United Kingdom MCML – Munich Center for Machine Learning Munich Germany MDSI – Munich Data Science Institute Munich Germany

Imbuing machines with the ability to talk has been a longtime pursuit of artificial intelligence (AI) research. From the very beginning, the community has not only aimed to synthesise high-fidelity speech that accurately conveys the semantic meaning of an utterance, but also to colour it with inflections that cover the same range of affective expressions that humans are capable of. After many years of research, it appears that we are on the cusp of achieving this when it comes to single, isolated utterances. This unveils an abundance of potential avenues to explore when it comes to combining these single utterances with the aim of synthesising more complex, longer-term behaviours. In the present chapter, we outline the methodological advances that brought us so far and sketch out the ongoing efforts to reach that coveted next level of artificial expressivity. We also discuss the societal implications coupled with rapidly advancing expressive speech synthesis (ESS) technology and highlight ways to mitigate those risks and ensure the alignment of ESS capabilities with ethical norms. Copyright © 2024, The Authors. All rights reserved.

关键词： Speech synthesis

The Implicit Regularization of Dynamical Stability in Stochastic Gradient Descent

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Wu, Lei Su, Weijie J. School of Mathematical Sciences Peking University Beijing China Center for Machine Learning Research Peking University Beijing China Wharton Statistics and Data Science Department University of Pennsylvania Philadelphia United States

In this paper, we study the implicit regularization of stochastic gradient descent (SGD) through the lens of dynamical stability (Wu et al., 2018). We start by revising existing stability analyses of SGD, showing how the Frobenius norm and trace of Hessian relate to different notions of stability. Notably, if a global minimum is linearly stable for SGD, then the trace of Hessian must be less than or equal to 2/η, where η denotes the learning rate. By contrast, for gradient descent (GD), the stability imposes a similar constraint but only on the largest eigenvalue of Hessian. We then turn to analyze the generalization properties of these stable minima, focusing specifically on two-layer ReLU networks and diagonal linear networks. Notably, we establish the equivalence between these metrics of sharpness and certain parameter norms for the two models, which allows us to show that the stable minima of SGD provably generalize well. By contrast, the stability-induced regularization of GD is provably too weak to ensure satisfactory generalization. This discrepancy provides an explanation of why SGD often generalizes better than GD. Note that the learning rate (LR) plays a pivotal role in the strength of stability-induced regularization. As the LR increases, the regularization effect becomes more pronounced, elucidating why SGD with a larger LR consistently demonstrates superior generalization capabilities. Additionally, numerical experiments are provided to support our theoretical findings. © 2023, CC BY-SA.

关键词： Stability

On the Generalization of Representation Uncertainty in Earth Observation

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Kondylatos, Spyros Bountos, Nikolaos Ioannis Michail, Dimitrios Zhu, Xiao Xiang Camps-Valls, Gustau Papoutsis, Ioannis Orion Lab National Observatory of Athens & National Technical University of Athens Greece Universitat de València Spain Harokopio University of Athens Greece Data Science in Earth Observation Technical University of Munich Germany Munich Center for Machine Learning Germany

Recent advances in Computer Vision have introduced the concept of pretrained representation uncertainty, enabling zero-shot uncertainty estimation. This holds significant potential for Earth Observation (EO), where trustworthiness is critical, yet the complexity of EO data poses challenges to uncertainty-aware methods. In this work, we investigate the generalization of representation uncertainty in EO, considering the domain’s unique semantic characteristics. We pretrain uncertainties on large EO datasets and propose an evaluation framework to assess their zero-shot performance in multi-label classification and segmentation EO tasks. Our findings reveal that, unlike uncertainties pretrained on natural images, EO-pretraining exhibits strong generalization across unseen EO domains, geographic locations, and target granularities, while maintaining sensitivity to variations in ground sampling distance. We demonstrate the practical utility of pretrained uncertainties showcasing their alignment with task-specific uncertainties in downstream tasks, their sensitivity to real-world EO image noise, and their ability to generate spatial uncertainty estimates out-of-the-box. Initiating the discussion on representation uncertainty in EO, our study provides insights into its strengths and limitations, paving the way for future research in the field. Code and weights are available at: https://***/Orion-AILab/EOUncertaintyGeneralization. © 2025, CC BY.

关键词： Semantic Segmentation

HUMAN-IN-THE-LOOP: TOWARDS LABEL EMBEDDINGS FOR MEASURING CLASSIFICATION DIFFICULTY

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Hechinger, Katharina Zhu, Xiao Xiang Koller, Christoph Kauermann, Göran Department of Statistics Ludwig-Maximilians-University Munich Germany Chair of Data Science in Earth Observation Technical University of Munich Munich Germany Munich Center for Machine Learning Munich Germany German Aerospace Center Wessling Germany

Uncertainty in machine learning models is a timely and vast field of research. In supervised learning, uncertainty can already occur in the first stage of the training process, the annotation phase. This scenario is particularly evident when some instances cannot be definitively classified. In other words, there is inevitable ambiguity in the annotation step and hence, not necessarily a single "ground truth" associated with each instance. The main idea of this work is to drop the assumption of a ground truth label and instead embed the annotations into a multidimensional space. This embedding is derived from the empirical distribution of annotations in a Bayesian setup, modeled via a Dirichlet-Multinomial framework. We estimate the model parameters and posteriors using a stochastic Expectation Maximisation algorithm with Markov Chain Monte Carlo steps. The methods developed in this paper readily extend to various situations where multiple annotators independently label instances. To showcase the generality of the proposed approach, we apply our approach to three benchmark datasets for image classification and Natural Language Inference, where multiple annotations per instance are available. Besides the embeddings, we can investigate the resulting correlation matrices, which reflect the semantic similarities of the original classes very well for all three exemplary datasets. Copyright © 2023, The Authors. All rights reserved.

关键词： Classification (of information)

machine learning With data Assimilation and Uncertainty Quantification for Dynamical Systems:A Review

学校读者我要写书评

暂无评论

IEEE/CAA Journal of Automatica Sinica 2023年第6期10卷 1361-1387页

作者： Sibo Cheng César Quilodrán-Casas Said Ouala Alban Farchi Che Liu Pierre Tandeo Ronan Fablet Didier Lucor Bertrand Iooss Julien Brajard Dunhui Xiao Tijana Janjic Weiping Ding Yike Guo Alberto Carrassi Marc Bocquet Rossella Arcucci Data Science Institute Department of ComputingImperial College LondonSW72AZ London Department of Earth Science and Engineering Imperial College LondonSW72AZ London Department of Computer Science and Engineering Hong Kong University of Science and TechnologyHong Kong 999077China the IMT Atlantique Lab-STICCUMR CNRS 6285France and OdysseyInria/IMTFrance.P.Tandeo is also with RIKEN Center for Computational ScienceKobeJapan the CEREA École des Ponts and EDF R&Dîle-de-FranceFrance the Laboratoire Interdisciplinaire des Sciences du Numérique CNRSParis-Saclay UniversityF-91403OrsayFrance the Electricitéde France(EDF) 78401 ChatouFranceInstitut de Mathématiques de Toulouse31062 ToulouseFrance and SINCLAIR AI LabSaclayFrance the Sorbonne University ParisFranceand also with Nansen Environmental and Remote Sensing Center(NERSC)BergenNorway the School of Mathematical Sciences Tongji UniversityShanghai 200092China the Mathematical Institute for Machine Learning and Data Science KU Eichstaett-IngolstadtBavariaGermany the School of Information Science and Technology Nantong UniversityNantong 226019China the Department of Physics and Astronomy“Augusto Righi” University of Bologna40124 BolognaItaly

data assimilation(DA)and uncertainty quantification(UQ)are extensively used in analysing and reducing error propagation in high-dimensional spatial-temporal *** applications span from computational fluid dynamics(CFD)to geoscience and climate ***,much effort has been given in combining DA,UQ and machine learning(ML)*** research efforts seek to address some critical challenges in high-dimensional dynamical systems,including but not limited to dynamical system identification,reduced order surrogate modelling,error covariance specification and model error correction.A large number of developed techniques and methodologies exhibit a broad applicability across numerous domains,resulting in the necessity for a comprehensive *** paper provides the first overview of state-of-the-art researches in this interdisciplinary field,covering a wide range of *** review is aimed at ML scientists who attempt to apply DA and UQ techniques to improve the accuracy and the interpretability of their models,but also at DA and UQ experts who intend to integrate cutting-edge ML approaches to their ***,this article has a special focus on how ML methods can overcome the existing limits of DA and UQ,and vice *** exciting perspectives of this rapidly developing research field are also *** Terms-data assimilation(DA),deep learning,machine learning(ML),reduced-order-modelling,uncertainty quantification(UQ).

关键词： assimilation overcome apply

SHADE: Deep Density-based Clustering

学校读者我要写书评

暂无评论

SHADE: Deep Density-based Clustering

IEEE International Conference on data Mining (ICDM)

作者： Anna Beer Pascal Weber Lukas Miklautz Collin Leiber Walid Durani Christian Böhm Claudia Plant Faculty of Computer Science University of Vienna Vienna Austria UniVie Doctoral School Computer Science Vienna Austria Database Systems and Data Mining LMU Munich Munich Germany Munich Center for Machine Learning Munich Germany ds: UniVie University of Vienna Vienna Austria

ISBN: (数字)9798331506681

ISBN: (纸本)9798331506698

Detecting arbitrarily shaped clusters in high-dimensional noisy data is challenging for current clustering methods. We introduce SHADE, the first deep clustering algorithm that incorporates density-connectivity into its loss function. Similar to existing deep clustering algorithms, SHADE supports high-dimensional and large data sets with the expressive power of a deep autoencoder. In contrast to most existing deep clustering methods that rely on a centroid-based clustering objective, SHADE incorporates a novel loss function that captures density-connectivity. It thereby learns a representation that enhances the separation of density-connected clusters. SHADE detects a stable clustering and noise points fully automatically without any user input. It outperforms existing methods in clustering quality, especially on data that contain non-Gaussian clusters, such as video data. Moreover, the embedded space of SHADE is suitable for visualization and interpretation of the clustering results as the individual shapes of the clusters are preserved.

关键词： data analysis Shape Clustering methods Noise Autoencoders data visualization Clustering algorithms Noise measurement data mining

Sampling Strategies for Compressive Imaging Under Statistical Noise

学校读者我要写书评

暂无评论

Sampling Strategies for Compressive Imaging Under Statistica...

2023 International Conference on Sampling Theory and Applications, SampTA 2023

作者： Hoppe, Frederik Krahmer, Felix Verdun, Claudio Mayrink Menzel, Marion I. Rauhut, Holger Rwth Aachen University Chair of Mathematics of Information Processing Aachen Germany Technical University of Munich Department of Mathematics Munich Germany Munich Center for Machine Learning Munich Germany Technical University of Munich Munich Data Science Institute Germany Technische Hochschule Ingolstadt AImotion Bavaria Ingolstadt Germany Technical University of Munich Department of Physics Garching Germany Ge Healthcare Munich Germany

ISBN: (纸本)9798350328851

Most of the compressive sensing literature in signal processing assumes that the noise present in the measurement has an adversarial nature, i.e., it is bounded in a certain norm. At the same time, the randomization introduced in the sampling scheme usually assumes an i.i.d. model where rows are sampled with replacement. In this case, if a sample is measured a second time, it does not add additional information. For many applications, where the statistical noise model is a more accurate one, this is not true anymore since a second noisy sample comes with an independent realization of the noise, so there is a fundamental difference between sampling with and without replacement. Therefore, a more careful analysis must be performed. In this short note, we illustrate how one can mathematically transition between these two noise models. This transition gives rise to a weighted LASSO reconstruction method for sampling without replacement, which numerically improves the solution of high-dimensional compressive imaging problems. © 2023 IEEE.

关键词： Compressed sensing

Advancing OCT-Based Retinal Disease Classification with XLSTM: A Framework for Variable-Length Volume Processing

学校读者我要写书评

暂无评论

Advancing OCT-Based Retinal Disease Classification with XLST...

IEEE International Symposium on Biomedical Imaging

作者： Emese Sükei Marzieh Oghbaie Ursula Schmidt-Erfurth Günter Klambauer Hrvoje Bogunović Department of Ophthalmology OPTIMA Lab Medical University of Vienna Austria Institute of Artificial Intelligence Medical University of Vienna Center for Medical Data Science Austria LIT AI Lab Institute for Machine Learning Johannes Kepler University Austria NXAI GmbH Linz Austria

ISBN: (数字)9798331520526

ISBN: (纸本)9798331520533

This paper presents a method for retinal disease classification using optical coherence tomography (OCT) scans, specifically addressing the challenge of variable B-scan density across dataset volumes. Deep learning methods often struggle with fluctuations in the number of B-scans within OCT volumes, which can limit their effectiveness. To address this, we propose an extended long short-term memory (xLSTM) model capable of processing variable-length B-scan sequences, capturing spatial and contextual information across entire volumes regardless of slice count. Leveraging xLSTM's advanced memory structure, our method enhances classification accuracy while maintaining computational efficiency. We evaluated the xLSTM-based approach on public and private datasets, demonstrating its robustness against inter-study variability and domain shifts. Experimental results show that this approach performs on par with or surpasses state-of-the-art baselines and offers a scalable solution for accurate 3D volume classification.

关键词： Three-dimensional displays Accuracy Optical coherence tomography Pipelines Memory management Retina Robustness Long short term memory Diseases Biomedical imaging

RRSIS: Referring Remote Sensing Image Segmentation

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Yuan, Zhenghang Mou, Lichao Hua, Yuansheng Zhu, Xiao Xiang Data Science in Earth Observation Technical University of Munich Munich80333 Germany The College of Civil and Transportation Engineering Shenzhen University Shenzhen518060 China The Munich Center for Machine Learning Munich80333 Germany

Localizing desired objects from remote sensing images is of great use in practical applications. Referring image segmentation, which aims at segmenting out the objects to which a given expression refers, has been extensively studied in natural images. However, almost no research attention is given to this task of remote sensing imagery. Considering its potential for real-world applications, in this paper, we introduce referring remote sensing image segmentation (RRSIS) to fill in this gap and make some insightful explorations. Specifically, we create a new dataset, called RefSegRS, for this task, enabling us to evaluate different methods. Afterward, we benchmark referring image segmentation methods of natural images on the RefSegRS dataset and find that these models show limited efficacy in detecting small and scattered objects. To alleviate this issue, we propose a language-guided cross-scale enhancement (LGCE) module that utilizes linguistic features to adaptively enhance multi-scale visual features by integrating both deep and shallow features. The proposed dataset, benchmarking results, and the designed LGCE module provide insights into the design of a better RRSIS model. The dataset and code will be available at https://***/ai4eo/reasoning/rrsis. Copyright © 2023, The Authors. All rights reserved.

关键词： Image segmentation