检索结果-内蒙古大学图书馆

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Simantiraki, Olympia Markaki, Maria Pantazis, Yannis Institute of Applied and Computational Mathematics Foundation for Research and Technology Hellas Greece

ISBN: (纸本)9798350368741

Synthetic speech is occasionally required to be enhanced by a fixed amount before being presented to the listener. However, this approach neglects the diverse types and levels of noise, potentially resulting in either unintelligible or unpleasant speech. This paper proposes a dynamic speech enhancement approach, which generates speech while accounting for the presence of noise. Our methodology extends the WaveRNN vocoder by conditioning not only the speech mel-spectrograms but also the spectrogram from the background noise. In training, the target speech is tilted according to the listeners' preference and we further introduce an objective metric that aims at maximizing the spectro-temporal regions where target energy exceeds the masker energy. The generated speech was tested in speech-shaped noise at various noise levels. Our evaluation results showed that (a) in more adverse conditions, both a fixed-amount post-enhanced baseline system and the suggested dynamically-enhanced system performed equally well in terms of intelligibility and preference and (b) in the least noisy conditions, where the intelligibility scores of all models were nearly identical, listeners preferred the dynamically-enhanced speech. Our findings demonstrate and reinforce the benefits of using dynamic speech enhancement techniques in noisy environments. © 2025 IEEE.

关键词： dynamic adaptation noise-aware synthesis speech enhancement Speech synthesis

来源：评论

学校读者我要写书评

暂无评论

Sub-pixel Material Fraction Mapping in the UrbanScape using PRISMA Hyperspectral Imagery

Sub-pixel Material Fraction Mapping in the UrbanScape using ...

引用

2023 Joint Urban Remote Sensing Event, JURSE 2023

作者： Poursanidis, Dimitris Panagiotakis, Emmanouil Chrysoulakis, Nektarios Institute of Applied and Computational Mathematics Foundation for Research and Technology Hellas Remote Sensing Lab Crete Heraklion Greece

ISBN: (纸本)9781665493734

New spaceborne imaging spectrometers will provide great opportunities for mapping urban materials. We use the PRISMA Hyperspectral Imagery to estimate the fractional covers of key urban classes at two levels of details. Impervious, pervious and vegetation at level I where at level II we split the class Impervious to manmade surfaces (cement, asphalt, and metal roofs), the class pervious to soil and non-photosynthetic vegetation (NPV) and the class vegetation into and high and low canopy. The use of Random Forests Regression with synthetically mixed training data generated from the image-based spectral library was used for fraction mapping. Mean Absolute Error for all classes, at both levels, was lower than 18%. All classes at level I achieve error values close to 10% while at level II the majority achieves close to 11% with the lowest (5%) to be for the class "metals"and the highest (17%) to be at the class "soil"which has been overestimated across the urbanscape. We resume that the combination of RFR and synthetically mixed training data generated from the image-based library can allow a generalized urban fraction mapping across various cities of same typology and region when field-based data are absent. © 2023 IEEE.

关键词： Hyperspectral imaging

来源：评论

学校读者我要写书评

暂无评论

Inferring the Parameters of Ultrafast Lasers from Images with Surface Patterns 13

Inferring the Parameters of Ultrafast Lasers from Images wit...

引用

13th International Symposium on Image and Signal Processing and Analysis, ISPA 2023

作者： Litsas, Anastasis Fraggelakis, Fotis Tsibidis, George D. Stratakis, Emmanuel Pantazis, Yannis University of Crete Computer Science Department Heraklion Crete Greece Institute of Electronic Structure and Laser Foundation for Research and Technology - Hellas Heraklion Crete Greece Institute of Applied and Computational Mathematics Foundation for Research and Technology - Hellas Heraklion Crete Greece

ISBN: (纸本)9798350315363

In this paper, we explore the potential of deep learning techniques in the field of ultra-fast laser processing. More specifically, we trained convolutional neural networks on an in-house dataset with the aim of predicting the laser parameters from images which capture the morphological characteristics of the material surface after being laser fabricated. There are several challenges related with the collected dataset stemming from stochastic variations of the material and the laser, the sampling scheme and the overall low sample size. Therefore, to further test our models we additionally construct a synthetic dataset. The results on the synthetic dataset are almost perfect showing that the proposed predictive models have the capacity to learn the assigned tasks. As expected, the predictive performance decreases when the real dataset is utilized. Nevertheless, we show that both the accuracy in the classification task and the mean square error in the majority of the regression tasks are satisfactory (e.g., classification accuracy drops from 99.9% to 94.8%). © 2023 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Physics-Informed Deep Learning Approach for Reintroducing Atomic Detail in Coarse-Grained Configurations of Multiple Poly(lactic acid) Stereoisomers (vol 64, pg 1853, 2024)

引用

JOURNAL OF CHEMICAL INFORMATION AND MODELING 2024年第17期64卷 6926-6926页

作者： Christofi, Eleftherios Bacova, Petra Harmandaris, Vagelis A. Computation-based Science and Technology Research Center The Cyprus Institute Nicosia 2121 Cyprus Departamento de Ciencia de los Materiales e Ingeniería Metalúrgica y Química Inorgánica Facultad de Ciencias IMEYMAT Campus Universitario Río San Pedro s/n. Puerto Real Cádiz 11510 Spain Department of Mathematics and Applied Mathematics University of Crete Heraklion GR-71110 Greece Institute of Applied and Computational Mathematics Foundation for Research and Technology - Hellas Heraklion GR-71110 Crete Greece

Multiscale modeling of complex molecular systems, such as macromolecules, encompasses methods that combine information from fine and coarse representations of molecules to capture material properties over a wide range of spatiotemporal scales. Being able to exchange information between different levels of resolution is essential for the effective transfer of this information. The inverse problem of reintroducing atomistic degrees of freedom in coarse-grained (CG) molecular configurations is particularly challenging as, from a mathematical point of view, it is an ill-posed problem; the forward mapping from the atomistic to the CG description is typically defined via a deterministic operator (“one-to-one” problem), whereas the reversed mapping from the CG to the atomistic model refers to creating one representative configuration out of many possible ones (“one-to-many” problem). Most of the backmapping methods proposed so far balance accuracy, efficiency, and general applicability. This is particularly important for macromolecular systems with different types of isomers, i.e., molecules that have the same molecular formula and sequence of bonded atoms (constitution) but differ in the three-dimensional configurations of their atoms in space. Here, we introduce a versatile deep learning approach for backmapping multicomponent CG macromolecules with chiral centers, trained to learn structural correlations between polymer configurations at the atomistic level and their corresponding CG descriptions. This method is intended to be simple and flexible while presenting a generic solution for resolution transformation. In addition, the method is aimed to respect the structural features of the molecule, such as local packing, capturing therefore the physical properties of the material. As an illustrative example, we apply the model on linear poly(lactic acid) (PLA) in melt, which is one of the most popular biodegradable polymers. The framework is tested on a number of model sys

关键词： Biopolymers Copolymers Monomers Organic polymers Plastics

来源：评论

学校读者我要写书评

暂无评论

Analysis Between Cultural Heritage, Human Settlements and Landscapes Using Earth Observation and Archaeological Data 3rd

Analysis Between Cultural Heritage, Human Settlements and La...

引用

3rd International Conference on Trandisciplinary Multispectral Modelling and Cooperation for the Preservation of Cultural Heritage, TMM_CH 2023

作者： Alexandrakis, George Kozyrakis, Georgios V. Paxakis, Nikos Parasyris, Antonios Vythoulka, Anastasia Kampanis, Nikolaos A. Moropoulou, Antonia Coastal Research Lab Institute of Applied and Computational Mathematics Foundation for Research & Technology-Hellas Heraklion Greece Lab of Materials Science and Engineering School of Chemical Engineering National Technical University of Athens Athens Greece

ISBN: (纸本)9783031422997

Geological and environmental conditions that influence local topography also affect indirectly the location of human settlement dynamics. Understanding those relationships plays an important role in environmental and cultural heritage research related to the evolution of settlement dynamics. This work aims to present preliminary results of a study of the patterns of human settlements, based on known archaeological sites at four small islands of Dodecanese, Greece, and to generate hypotheses about the relations of cultural heritage patterns with the landscape. To that end, a Web-GIS database was created, which was fed with topographic, geomorphological data and Earth Observation data. Geomorphological analysis, derived from earth observation products, can provide useful estimations into the processes that shape landscapes and insight into the location and evolution of settlements. The analysis includes a series of different data correlation, from geomorphologic to socioeconomic, integrated by an indicator analysis. A series of thematic maps were developed to interpret why areas were selected to host settlements. Using the database, a set of indexes have been applied, that included exposure and vulnerability indices for the inland and coastal areas for specific sites. The areas under investigation present different characteristics in their geomorphological but also their societal evolution. Geomorphological data was further analysed in a ternary diagram that indicated the relative influence of each of the parameters in each area. From the diagram it is seen that the locations of human activities are strongly affected by environmental factors. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.

关键词： Location

来源：评论

学校读者我要写书评

暂无评论

Efficient discrimination between biological populations via neural-based estimation of RÉnyi divergence 4

Efficient discrimination between biological populations via ...

引用

4th International Conference on Uncertainty Quantification in computational Sciences and Engineering, UNCECOMP 2021

作者： Tsourtis, Anastasios Papoutsoglou, Georgios Pantazis, Yannis Institute of Applied and Computational Mathematics Foundation for Research and Technology Hellas Greece

The advent of single-cell or single-molecule sampling techniques allowed the study of abnormalities in small subsets of cell populations as well as subtle differences of evolving biological phenomena. A major challenge in this effort is the quantification of statistical differences between the probability distributions of measured quantities;particularly when small perturbations or small distributional changes need to be detected. Here, we propose to use as a discriminative tool the Rényi divergence whose key advantage is its ability to highlight differences between probability tails. In addition, we describe an algorithm which is based on a variational representation formula for the Rényi divergence and implicitly estimates it by solving an optimization problem. We evaluate the discrimination performance on both synthetic and real datasets. The proposed algorithm is able to detect distributional differences which are below 0.5% and quantify the trade-off between number of samples and neural network complexity. The comparison with existing density ratio approaches reveals that the proposed method is significantly better when the dimension of the data is moderately high (e.g., larger than 10). © 2021 UNCECOMP Proceedings. All rights reserved.

关键词： Probability distributions

来源：评论

学校读者我要写书评

暂无评论

Dynamic Speech Generation to Enhance Intelligibility in Noisy Environments

Dynamic Speech Generation to Enhance Intelligibility in Nois...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Olympia Simantiraki Maria Markaki Yannis Pantazis Institute of Applied and Computational Mathematics Foundation for Research and Technology Greece

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Synthetic speech is occasionally required to be enhanced by a fixed amount before being presented to the listener. However, this approach neglects the diverse types and levels of noise, potentially resulting in either unintelligible or unpleasant speech. This paper proposes a dynamic speech enhancement approach, which generates speech while accounting for the presence of noise. Our methodology extends the WaveRNN vocoder by conditioning not only the speech mel-spectrograms but also the spectrogram from the background noise. In training, the target speech is tilted according to the listeners’ preference and we further introduce an objective metric that aims at maximizing the spectro-temporal regions where target energy exceeds the masker energy. The generated speech was tested in speech-shaped noise at various noise levels. Our evaluation results showed that (a) in more adverse conditions, both a fixed-amount post-enhanced baseline system and the suggested dynamically-enhanced system performed equally well in terms of intelligibility and preference and (b) in the least noisy conditions, where the intelligibility scores of all models were nearly identical, listeners preferred the dynamically-enhanced speech. Our findings demonstrate and reinforce the benefits of using dynamic speech enhancement techniques in noisy environments.

关键词： Training Adaptation models Vocoders Virtual assistants Noise Speech enhancement Acoustics Noise measurement Speech synthesis Spectrogram

来源：评论

学校读者我要写书评

暂无评论

Initial development of the urbisphere urban hyperspectral library: Berlin, Germany

Initial development of the urbisphere urban hyperspectral li...

引用

2023 Joint Urban Remote Sensing Event, JURSE 2023

作者： Lantzanakis, Giannis Tsirantonakis, Dimitris Chrysoulakis, Nektarios Grimmond, Sue Christen, Andreas Birkmann, Joern Remote Sensing Lab Foundation for Research and Technology Hellas Institute of Applied and Computational Mathematics Heraklion Greece University of Reading Department of Meteorology Reading United Kingdom University of Stuttgart Institute of Spatial and Regional Planning Stuttgart Germany

ISBN: (纸本)9781665493734

The future sustainability of cities is linked directly with their adaptive capacity in response to both ongoing urbanization, weather extremes and climate change. Many possible actions are related to urban materials and their thermal and radiative properties. To explore alternatives using modelling and to monitor applications the thermal and radiative properties of materials are critical. However, there is a lack of hyperspectral data for a wide range of common materials. A Spectral Evolution RS-3500 spectroradiometer is used to measure 2151 spectral (1 nm) bands between 350 to 2500 nm in four representative neighborhoods of Berlin (Charlottenburg, Neukolln, Lichtenberg and Mitte) during August 2022. The data are processed to contribute 284 hyperspectral measurements of surfaces into the urbisphere hyperspectral library. © 2023 IEEE.

关键词： Climate change

来源：评论

学校读者我要写书评

暂无评论

On the Γ-convergence of the Allen-Cahn functional with boundary conditions

arXiv

引用

arXiv 2023年

作者： Gazoulis, Dimitrios Department of Mathematics and Applied Mathematics University of Crete Heraklion70013 Greece Institute of Applied and Computational Mathematics Foundation for Research and Technology-Hellas Greece

We study minimizers of the Allen-Cahn system. We consider the "-energy functional with Dirichlet values and we establish the Γ-limit. The minimizers of the limiting functional are closely related to minimizing partitions of the domain. Finally, utilizing that the triod and the straight line are the only minimal cones in the plane together with regularity results for minimal curves, we determine the precise structure of the minimizers of the limiting functional, and thus the limit of minimizers of the "-energy functional as " → 0. © 2023, CC BY.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Sub-pixel Material Fraction Mapping in the UrbanScape using PRISMA Hyperspectral Imagery

Sub-pixel Material Fraction Mapping in the UrbanScape using ...

引用

Urban Remote Sensing Joint Event

作者： Dimitris Poursanidis Emmanouil Panagiotakis Nektarios Chrysoulakis Remote Sensing Lab Institute of Applied and Computational Mathematics Foundation for Research and Technology Hellas Heraklion Crete Greece

New spaceborne imaging spectrometers will provide great opportunities for mapping urban materials. We use the PRISMA Hyperspectral Imagery to estimate the fractional covers of key urban classes at two levels of details. Impervious, pervious and vegetation at level I where at level II we split the class Impervious to manmade surfaces (cement, asphalt, and metal roofs), the class pervious to soil and non-photosynthetic vegetation (NPV) and the class vegetation into and high and low canopy. The use of Random Forests Regression with synthetically mixed training data generated from the image-based spectral library was used for fraction mapping. Mean Absolute Error for all classes, at both levels, was lower than 18%. All classes at level I achieve error values close to 10% while at level II the majority achieves close to 11% with the lowest (5%) to be for the class "metals" and the highest (17%) to be at the class "soil" which has been overestimated across the urbanscape. We resume that the combination of RFR and synthetically mixed training data generated from the image-based library can allow a generalized urban fraction mapping across various cities of same typology and region when field-based data are absent.

关键词：

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：