检索结果-内蒙古大学图书馆

Synthesis Speech Based Data Augmentation for Low Resource Children ASR 23rd

学校读者我要写书评

暂无评论

Synthesis Speech Based Data Augmentation for Low Resource Ch...

23rd International Conference on Speech and computer, SPECOM 2021

作者： Kadyan, Virender Kathania, Hemant Govil, Prajjval Kurimo, Mikko Speech and Language Research Centre School of Computer Science University of Petroleum and Energy Studies Dehradun Uttarakhand India Department of Electronics and Communication Engineering National Institute of Sikkim Ravangla India Department of Signal Processing and Acoustics Aalto University Espoo Finland

ISBN: (纸本)9783030878016

Successful speech recognition for children requires large training data with sufficient speaker variability. The collection of such a training database of children’s voices is challenging and very expensive for zero/low resource language like Punjabi. In this paper, the data scarcity issue of the low resourced language Punjabi is addressed through two levels of augmentation. The original training corpus is first augmented by modifying the prosody parameters for pitch and speaking rate. Our results show that the augmentation improves the system performance over the baseline system. Then the augmented data combined with original data and used to train the TTS system to generate synthesis data and extended dataset is further used for augmented by generating children’s utterances using text-to-speech synthesis and sampling the language model with methods that increase the acoustic and lexical diversity. The final speech recognition performance indicates a relative improvement of 50.10% with acoustic and 57.40% with language diversity based augmentation in comparison to that of the baseline system respectively. © 2021, Springer Nature Switzerland AG.

关键词： Speech synthesis

SurgicalGaussian: Deformable 3D Gaussians for High-Fidelity Surgical Scene Reconstruction

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Xie, Weixing Yao, Junfeng Cao, Xianpeng Lin, Qiqin Tang, Zerui Dong, Xiao Guo, Xiaohu Center for Digital Media Computing School of Film School of Informatics Xiamen University Xiamen China National Institute for Data Science in Health and Medicine Xiamen University China Key Laboratory of Digital Protection and Intelligent Processing of Intangible Cultural Heritage of Fujian and Taiwan Ministry of Culture and Tourism China Guangdong Provincial Key Laboratory IRADS Department of Computer Science BNU-HKBU United International College Zhuhai China Department of Computer Science The University of Texas at Dallas Dallas United States

Dynamic reconstruction of deformable tissues in endoscopic video is a key technology for robot-assisted surgery. Recent reconstruction methods based on neural radiance fields (NeRFs) have achieved remarkable results in the reconstruction of surgical scenes. However, based on implicit representation, NeRFs struggle to capture the intricate details of objects in the scene and cannot achieve real-time rendering. In addition, restricted single view perception and occluded instruments also propose special challenges in surgical scene reconstruction. To address these issues, we develop SurgicalGaussian, a deformable 3D Gaussian Splatting method to model dynamic surgical scenes. Our approach models the spatio-temporal features of soft tissues at each time stamp via a forward-mapping deformation MLP and regularization to constrain local 3D Gaussians to comply with consistent movement. With the depth initialization strategy and tool mask-guided training, our method can remove surgical instruments and reconstruct high-fidelity surgical scenes. Through experiments on various surgical videos, our network outperforms existing method on many aspects, including rendering quality, rendering speed and GPU usage. The project page can be found at https://***. © 2024, CC BY-SA.

关键词： Tissue

r-Minimal Codes with Respect to Rank Metric

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Xu, Yang Kan, Haibin Han, Guangyue Shanghai Key Laboratory of Intelligent Information Processing School of Computer Science Fudan University Shanghai200433 China Shanghai Engineering Research Center of Blockchain Shanghai200433 China Yiwu Research Institute Fudan University Zhejiang Yiwu City322000 China Department of Mathematics Faculty of Science The University of Hong Kong Pokfulam Road Hong Kong

In this paper, we propose and study r-minimal codes, a natural extension of minimal codes which have been extensively studied with respect to Hamming metric, rank metric and sum-rank metric. We first propose r-minimal codes in a general setting where the ambient space is a finite dimensional left module over a division ring and is supported on a lattice. We characterize minimal subcodes and r-minimal codes, derive a general singleton bound, and give existence results for r-minimal codes by using combinatorial arguments. We then consider r-minimal rank metric codes over a field extension E/F of degree m, where E can be infinite. We characterize these codes in terms of cutting r-blocking sets, generalized rank weights of the codes and those of the dual codes, and classify codes whose r-dimensional subcodes have constant rank support weight. Next, with the help of the evasiveness property of cutting r-blocking sets and some upper bounds for the dimensions of evasive subspaces, we derive several lower and upper bounds for the minimal length of r-minimal codes. Furthermore, when E is finite, we establish a general upper bound which generalizes and improves the counterpart for minimal codes in the literature. As a corollary, we show that if m = 3, then for any k > 2, the minimal length of k-dimensional minimal codes is equal to 2k. To the best of our knowledge, when m > 3, there was no known explicit formula for the minimal length of k-dimensional minimal codes for arbitrary k in the literature. © 2024, CC BY.

关键词： Turbo codes

image Reconstruction with B0 Inhomogeneity using an Interpretable Deep Unrolled Network on an Open-bore MRI-Linac

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Shan, Shanshan Gao, Yang Waddington, David E.J. Chen, Hongli Whelan, Brendan Liu, Paul Z.Y. Wang, Yaohui Liu, Chunyi Gan, Hongping Gao, Mingyuan Liu, Feng Soochow University Suzhou215123 China Image X Institute Sydney School of Health Sciences Faculty of Medicine and Health University of Sydney Sydney2015 Australia Department of Medical Physics Ingham Institute of Applied Medical Research Liverpool2170 Australia School of Computer Science and Engineering Central South University Changsha410083 China School of Electrical Engineering and Computer Science University of Queensland Brisbane4072 Australia Institute of Electrical Engineering Chinese Academy of Science Beijing100190 China School of Software Northwestern Polytechnical University Taicang215400 China

MRI-Linac systems require fast image reconstruction with high geometric fidelity to localize and track tumours for radiotherapy treatments. However, B0 field inhomogeneity distortions and slow MR acquisition potentially limit the quality of the image guidance and tumour treatments. In this study, we develop an interpretable unrolled network, referred to as RebinNet, to reconstruct distortion-free images from B0 inhomogeneity-corrupted k-space for fast MRI-guided radiotherapy applications. RebinNet includes convolutional neural network (CNN) blocks to perform image regularizations and nonuniform fast Fourier Transform (NUFFT) modules to incorporate B0 inhomogeneity information. The RebinNet was trained on a publicly available MR dataset from eleven healthy volunteers for both fully sampled and subsampled acquisitions. Grid phantom and human brain images acquired from an open-bore 1T MRI-Linac scanner were used to evaluate the performance of the proposed network. The RebinNet was compared with the conventional regularization algorithm and our recently developed UnUNet method in terms of root mean squared error (RMSE), structural similarity (SSIM), residual distortions, and computation time. Imaging results demonstrated that the RebinNet reconstructed images with lowest RMSE (0.92) at four-time acceleration for simulated brain images. The RebinNet could better preserve structural details and substantially improve the computational efficiency (ten-fold faster) compared to the conventional regularization methods, and had better generalization ability than the UnUNet method. The proposed RebinNet can achieve rapid image reconstruction and overcome the B0 inhomogeneity distortions simultaneously, which would facilitate accurate and fast image guidance in radiotherapy treatments. © 2024, CC BY-NC-ND.

关键词： Magnetic resonance imaging

Field Measurements and Energy Performance Modelling of Water Side Free Cooling System for Datacenters

学校读者我要写书评

暂无评论

Field Measurements and Energy Performance Modelling of Water...

East-West Design & Test Symposium (EWDTS)

作者： Hande İlkor Gül Nihal Güğül Mustafa Kemerci Tugğba Güneş Department of Computer Engineering Faculty of Technology Institute of Science Selcuk University Konya Turkey Department of Computer Engineering Faculty of Technology Selcuk University Konya Turkey Information Processing Department Republic of Turkey Ministry of Environment Urbanization and Climate Change General Directorate of Land Registry and Cadastre Ankara Türkiye Cadastre Department Republic of Turkey Ministry of Environment Urbanization and Climate Change General Directorate of Land Registry and Cadastre Ankara Türkiye

ISBN: (数字)9798331515768

ISBN: (纸本)9798331515775

In recent years, number of free cooling system (FCS) applied in datacenters (DC) is increasing due to the provided significant decrease in the electricity consumption and short payback period. However, since FCS is a partially new technology, energy savings potential of FCS has not been sufficiently investigated using real time data series in the literature. In this study, the field measurements and energy performance analysis were performed in an operational DC with water-side free cooling, located in Ankara, Türkiye. Mathematical model of FCS is developed using electricity consumption data of IT devices and FCS, and climate data of Ankara using MS Excel. Electricity consumption of FCS and the PUE of the DC in case of installation of the existing DC in six different climate regions of Turkey were calculated using the developed model. In case of using the FCS, PUE value is obtained less than 1.1 for each location. The result shows that, very low temperatures are not required for the FCS, and the important factor is the number of hours that is lower than the indoor temperature of DC. In case of installation of the DC in Antalya and Şanlıurfa, which are among hottest cities of Türkiye, the electricity consumption of the FCS increased by 8 % and 9 %. However, since the energy consumption of FCS is still much lower than the IT consumption compared to the old type cooling systems, the use of the FCS provided adequate saving.

关键词： Temperature measurement Performance evaluation Analytical models Temperature dependence Cooling Electricity Atmospheric modeling Energy measurement Mathematical models Meteorology

MG-SLAM: Structure Gaussian Splatting SLAM with Manhattan World Hypothesis

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Liu, Shuhong Deng, Tianchen Zhou, Heng Li, Liuzhuozheng Wang, Hongyu Wang, Danwei Li, Mingrui Department of Information Science and Technology Department of Complexity Science and Engineering The University of Tokyo Tokyo113-8654 Japan Institute of Medical Robotics Department of Automation Shanghai Jiao Tong University Key Laboratory of System Control and Information Processing Ministry of Education Shanghai200240 China Department of Mechanical Engineering Columbia University New York10027 United States School of Electrical and Electronic Engineering Nanyang Technological University Singapore Department of Computer Science Dalian University of Technology Dalian116024 China

Gaussian Splatting SLAMs have made significant advancements in improving the efficiency and fidelity of real-time reconstructions. However, these systems often encounter incomplete reconstructions in complex indoor environments, characterized by substantial holes due to unobserved geometry caused by obstacles or limited view angles. To address this challenge, we present Manhattan Gaussian SLAM, an RGB-D system that leverages the Manhattan World hypothesis to enhance geometric accuracy and completeness. By seamlessly integrating fused line segments derived from structured scenes, our method ensures robust tracking in textureless indoor areas. Moreover, The extracted lines and planar surface assumption allow strategic interpolation of new Gaussians in regions of missing geometry, enabling efficient scene completion. Extensive experiments conducted on both synthetic and real-world scenes demonstrate that these advancements enable our method to achieve state-of-the-art performance, marking a substantial improvement in the capabilities of Gaussian SLAM systems. © 2024, CC BY.

关键词： Gaussian distribution

Embedding Space Augmentation for Weakly Supervised Learning in Whole-Slide images

学校读者我要写书评

暂无评论

Embedding Space Augmentation for Weakly Supervised Learning ...

IEEE International Symposium on Biomedical Imaging

作者： Imaad Zaffar Guillaume Jaume Nasir Rajpoot Faisal Mahmood Department of Computer Science University College London UK Department of Pathology Harvard Medical School Brigham and Women’s Hospital Boston MA USA Department of Pathology Harvard Medical School Massachusetts General Hospital Boston MA USA Cancer Program Broad Institute of Harvard and MIT Cambridge MA USA Data Science Program Dana-Farber Cancer Institute Boston MA USA Department of Computer Science Tissue Image Analytics Centre University of Warwick Coventry UK Department of Pathology University Hospitals Coventry and Warwickshire NHS Trust Coventry UK The Alan Turing Institute London UK

Multiple Instance Learning (MIL) is a widely employed framework for learning on gigapixel whole-slide images (WSIs) from WSI-level annotations. In most MIL-based analytical pipelines for WSI-level analysis, the WSIs are divided into patches, and deep features for patches (i.e., patch embeddings) are extracted prior to training to reduce the overall computational cost and cope with the GPUs’ limited RAM. Because of this bottleneck, incorporating patch-level data augmentations during training adds an extra computational burden. To overcome this limitation, we present EmbAugmenter, a data augmentation generative adversarial network (DA-GAN) that can synthesize data augmentations in the embedding space rather than in the pixel space, thereby significantly reducing the computational requirements. Experiments on the SICAPv2 dataset show that our approach outperforms MIL without augmentation and is on par with traditional patch-level augmentation for MIL training while being substantially faster.

关键词：

A perspective on Gaussian processes for Earth observation

学校读者我要写书评

暂无评论

National science Review 2019年第4期6卷 616-618页

作者： Gustau Camps-Valls Dino Sejdinovic Jakob Runge Markus Reichstein Image Processing Laboratory University of Valencia Department of Statistics University of Oxford German Aerospace Center Institute of Data Science Max Planck Institute for Biogeochemistry

INTRODUCTION Earth observation (EO) by airborne and satellite remote sensing and in situ observations plays a fundamental role in monitoring our planet. In the last decade, machine learning has attained outstanding results in the estimation of bio-geo-physical variables from the acquired images at local and global scales in a time-resolved manner. Gaussian processes (GPs)[1], as flexible

关键词： Earth Gaussian A perspective on Gaussian processes for Earth observation

Well-tuned simple nets excel on tabular datasets 21

学校读者我要写书评

暂无评论

Well-tuned simple nets excel on tabular datasets

Proceedings of the 35th International Conference on Neural Information processing Systems

作者： Arlind Kadra Marius Lindauer Frank Hutter Josif Grabocka Department of Computer Science University of Freiburg Institute for Information Processing Leibniz University Hannover Department of Computer Science University of Freiburg & Bosch Center for Artificial Intelligence

ISBN: (纸本)9781713845393

Tabular datasets are the last "unconquered castle" for deep learning, with traditional ML methods like Gradient-Boosted Decision Trees still performing strongly even against recent specialized neural architectures. In this paper, we hypothesize that the key to boosting the performance of neural networks lies in rethinking the joint and simultaneous application of a large set of modern regularization techniques. As a result, we propose regularizing plain Multilayer Perceptron (MLP) networks by searching for the optimal combination/cocktail of 13 regularization techniques for each dataset using a joint optimization over the decision on which regularizers to apply and their subsidiary *** empirically assess the impact of these regularization cocktails for MLPs in a large-scale empirical study comprising 40 tabular datasets and demonstrate that (i) well-regularized plain MLPs significantly outperform recent state-of-the-art specialized neural network architectures, and (ii) they even outperform strong traditional ML methods, such as XGBoost.

关键词：