检索结果-内蒙古大学图书馆

DeepCNN:Spectro-temporal feature representation for speech emotion recognition

学校读者我要写书评

暂无评论

CAAI Transactions on Intelligence Technology 2023年第2期8卷 401-417页

作者： Nasir Saleem Jiechao Gao Rizwana Irfan Ahmad Almadhor Hafiz Tayyab Rauf Yudong Zhang Seifedine Kadry Department of Electrical Engineering Faculty of Engineering and TechnologyGomal UniversityD.I.KhanPakistan Department of Computer Science University of VirginiaCharlottesvilleVirginiaUSA Department of Information Technology College of Computing and Information Technology at KhulaisUniversity of JeddahJeddahSaudi Arabia Department of Computer Engineering and Networks College of Computer and Information SciencesJouf UniversitySkaka AljoufSaudi Arabia Centre for Smart Systems AI and CybersecurityStaffordshire UniversityStoke‐on‐TrentUK School of Computing and Mathematical Sciences University of LeicesterLeicesterUK Department of Applied Data Science Noroff University CollegeKristiansandNorway Artificial Intelligence Research Center(AIRC) Ajman UniversityAjmanUnited Arab Emirates Department of Electrical and Computer Engineering Lebanese American UniversityByblosLebanon

Speech emotion recognition(SER)is an important research problem in human-computer interaction *** representation and extraction of features are significant challenges in SER *** the promising results of recent studies,they generally do not leverage progressive fusion techniques for effective feature representation and increasing receptive *** mitigate this problem,this article proposes DeepCNN,which is a fusion of spectral and temporal features of emotional speech by parallelising convolutional neural networks(CNNs)and a convolution layer-based *** parallel CNNs are applied to extract the spectral features(2D-CNN)and temporal features(1D-CNN)representations.A 2D-convolution layer-based transformer module extracts spectro-temporal features and concatenates them with features from parallel *** learnt low-level concatenated features are then applied to a deep framework of convolutional blocks,which retrieves high-level feature representation and subsequently categorises the emotional states using an attention gated recurrent unit and classification *** fusion technique results in a deeper hierarchical feature representation at a lower computational cost while simultaneously expanding the filter depth and reducing the feature *** Berlin database of Emotional Speech(EMO-BD)and Interactive Emotional Dyadic Motion Capture(IEMOCAP)datasets are used in experiments to recognise distinct speech *** efficient spectral and temporal feature representation,the proposed SER model achieves 94.2%accuracy for different emotions on the EMO-BD and 81.1%accuracy on the IEMOCAP dataset *** proposed SER system,DeepCNN,outperforms the baseline SER systems in terms of emotion recognition accuracy on the EMO-BD and IEMOCAP datasets.

关键词： decision making deep learning

LP-Gaussians: Learnable Parametric Gaussian Splatting for Efficient Dynamic Reconstruction of Single-View Scenes

学校读者我要写书评

暂无评论

LP-Gaussians: Learnable Parametric Gaussian Splatting for Ef...

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

作者： Shaoqi Wu Weixing Xie Youhong Peng Jinwen Li Jiawei Yao Junfeng Yao Center for Digital Media Computing School of Film School of Informatics Xiamen University 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

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

With the popularity of short video platforms, the number of single-view videos has increased significantly. Existing NeRF-based methods can reconstruct dynamic scenes in a single-view setting, but slow rendering speed and low rendering quality limit their practical applications. To address these challenges, we propose a fast single-view scenes reconstruction framework based on 3D Gaussian Splatting. Our method uses point clouds obtained with depth priors as the Gaussian initialization and introduces learnable parametric functions to model the time-dependent deformation of Gaussians. The explicit deformation modeling for Gaussians significantly reduces training and rendering time. Furthermore, to improve the rendering quality of challenging areas, we adopt an adaptive sampling strategy to densify Gaussians. For occlusion problems from single-view videos, we design a smooth loss function to restore the color of the occluded areas. Experimental results demonstrate that our method significantly reduces training time, enhances rendering quality, and accelerates rendering speed. Project page: https://***/LPGaussians.

关键词： Training Point cloud compression Deformable models Adaptation models Three-dimensional displays Deformation Streaming media Signal processing Rendering (computer graphics) Speech processing

Electrocardiogram classification and visual diagnosis of atrial fibrillation with DenseECG

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Chen, Dacheng Li, Dan Xu, Xiuqin Yang, Ruizhi Ng, See-Kiong Institute of Data Science National University of Singapore Singapore School of Computing National University of Singapore Singapore School of Computing and Institute of Data Science National University of Singapore Singapore

Atrial Fibrillation (AF) is a common cardiac arrhythmia affecting a large number of people around the world. If left undetected, it will develop into chronic disability or even early mortality. However, patients who have this problem can barely feel its presence, especially in its early stage. A noninvasive, automatic and effective detection method is therefore needed to help early detection so that medical intervention can be implemented in time to prevent its progression. Electrocardiogram (ECG), which records the electrical activities of the heart, has been widely used for detecting the presence of AF. However, due to the subtle patterns of AF, the performance of detection models have largely depended on complicated data pre-processing and expertly engineered features. In our work, we developed DenseECG, an end-to-end model based on a 5 layers 1D densely connected convolutional neural network. We trained our model using the publicly available dataset from 2017 PhysioNet computing in Cardiology(CinC) Challenge containing 8528 single-lead ECG recordings of short term heart rhythms (9-61s). Our trained model was able to outperform the other state-of-the-art AF detection models on this dataset without complicated data pre-processing and expert-supervised feature engineering. In addition, we address the need for explainability and traceability of machine learning models for medical diagnosis by plotting the class activation map (CAM) of an input ECG sequence with our trained AF detection model. CAM had been used to localize the discriminative regions of input images for image classification tasks. In this work, we adopted it to highlight the discriminative regions of the ECG data to visually unveil where our AF detection model was focusing on in classifying an input ECG sequence for AF. Furthermore, the DenseECG model is applied to another more imbalanced MIT-BTH dataset. Our proposed model with class weight significantly outperforms the state-of-the-art models. Copyr

关键词： Diseases

Voice Conversion Augmentation for Speaker Recognition on Defective datasets

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Tao, Ruijie Shi, Zhan Jiang, Yidi Liu, Tianchi Li, Haizhou Department of Electrical and Computer Engineering National University of Singapore Singapore119077 Singapore School of Data Science The Chinese University of Hong Kong Shenzhen518172 China Singapore Institute of Technology Institute for Infocomm Research A∗STAR Singapore138632 Singapore Guangdong Provincial Key Laboratory of Big Data Computing The Chinese University of Hong Kong Shenzhen518172 China Shenzhen Research Institute of Big data Shenzhen518172 China University of Bremen Bremen28359 Germany

Modern speaker recognition system relies on abundant and balanced datasets for classification training. However, diverse defective datasets, such as partially-labelled, small-scale, and imbalanced datasets, are common in real-world applications. Previous works usually studied specific solutions for each scenario from the algorithm perspective. However, the root cause of these problems lies in dataset imperfections. To address these challenges with a unified solution, we propose the Voice Conversion Augmentation (VCA) strategy to obtain pseudo speech from the training set. Furthermore, to guarantee generation quality, we designed the VCA-NN (nearest neighbours) strategy to select source speech from utterances that are close to the target speech in the representation space. Our experimental results on three created datasets demonstrated that VCA-NN effectively mitigates these dataset problems, which provides a new direction for handling the speaker recognition problems from the data aspect. © 2024, CC BY.

关键词： Classification (of information)

Deep Learning Approaches for Road Damage Detection Using YOLOv8 and Custom Convolutional Neural Network

学校读者我要写书评

暂无评论

Deep Learning Approaches for Road Damage Detection Using YOL...

Frontiers Technology of Information and Computer (ICFTIC), IEEE International Conference on

作者： Muhammad Waseem Khan Khalid Mahmood Israr Hussain Muhammad Talha Qureshi Hafiz Muhammad Sanaullah Badar Wu Gao Center for Bio-inspired System-on-Chip School of Computer Science Northwestern Poly technical University Xi'an P. R. China Graduate School of Intelligent Data Science National Yunlin University of Science and Technology Taiwan R.O.C Institute of Computing MNS University of Agriculture Multan Pakistan Henan University Kaifeng China

ISBN: (数字)9798331541750

ISBN: (纸本)9798331541767

The identification of road damage is deemed very important to the preservation of infrastructure. Since recent development in deep learning provide great potential approaches, in this paper, three deep learning models, namely EfficientNet, CNN and YOLOv8 are considered and their performance is evaluated on a dataset with 8,586 images with labelled road damages consisting of seven subclasses including longitudinal cracks and potholes. Besides, other issues like class imbalance and variabilities in the environment were studied and considered in its assessment based on the Models accuracy, precision, recall, and F1-score. With an accuracy of 64%, YOLOv8 is the most accurate model for identifying various types of damages, particularly in identifying multiple damages present in one or more types of damage, as was seen in the high accuracy rate of a split sample of 44 images of multiple damages. Our findings showcase YOLOv8 as a strong contender for real-time road damage detection with potential of enhancement with better augmentation and fine-tuning.

关键词： Deep learning Location awareness Accuracy Roads Image edge detection Transfer learning Real-time systems data models Convolutional neural networks Reliability

Hypothesis testing for differentially private linear regression 22

学校读者我要写书评

暂无评论

Hypothesis testing for differentially private linear regress...

Proceedings of the 36th International Conference on Neural Information Processing Systems

作者： Daniel Alabi Salil Vadhan Department of Computer Science and Data Science Institute Columbia University Harvard School of Engineering and Applied Sciences

ISBN: (纸本)9781713871088

In this work, we design differentially private hypothesis tests for the following problems in the general linear model: testing a linear relationship and testing for the presence of mixtures. The majority of our hypothesis tests are based on differentially private versions of the F-statistic for the general linear model framework, which are uniformly most powerful unbiased in the non-private setting. We also present another test for testing mixtures, based on the differentially private nonparametric tests of Couch, Kazan, Shi, Bray, and Groce (CCS 2019), which is especially suited for the small dataset regime. We show that the differentially private F-statistic converges to the asymptotic distribution of its non-private counterpart. As a corollary, the statistical power of the differentially private F-statistic converges to the statistical power of the non-private F-statistic. Through a suite of Monte Carlo based experiments, we show that our tests achieve desired significance levels and have a high power that approaches the power of the non-private tests as we increase sample sizes or the privacy-loss parameter. We also show when our tests outperform existing methods in the literature.

关键词：

Lost in UNet: Improving Infrared Small Target Detection by Underappreciated Local Features

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Quan, Wuzhou Zhao, Wei Wang, Weiming Xie, Haoran Wang, Fu Lee Wei, Mingqiang The School of Computer Science and Technology Nanjing University of Aeronautics and Astronautics Nanjing China The Shenzhen Research Institute Nanjing University of Aeronautics and Astronautics Shenzhen China The School of Science and Technology Hong Kong Metropolitan University Hong Kong The School of Data Science Lingnan University Hong Kong

Many targets are often very small in infrared images due to the long-distance imaging meachnism. UNet and its variants, as popular detection backbone networks, downsample the local features early and cause the irreversible loss of these local features, leading to both the missed and false detection of small targets in infrared images. We propose HintU, a novel network to recover the local features lost by various UNet-based methods for effective infrared small target detection. HintU has two key contributions. First, it introduces the "Hint" mechanism for the first time, i.e., leveraging the prior knowledge of target locations to highlight critical local features. Second, it improves the mainstream UNet-based architecture to preserve target pixels even after downsampling. HintU can shift the focus of various networks (e.g., vanilla UNet, UNet++, UIUNet, MiM+, and HCFNet) from the irrelevant background pixels to a more restricted area from the beginning. Experimental results on three datasets NUDT-SIRST, SIRSTv2 and IRSTD1K demonstrate that HintU enhances the performance of existing methods with only an additional 1.88 ms cost (on RTX Titan). Additionally, the explicit constraints of HintU enhance the generalization ability of UNet-based methods. Code is available at https://***/Wuzhou-Quan/HintU. Copyright © 2024, The Authors. All rights reserved.

关键词： Pixels

FCS-MPC with Computational Efficiency and dv/dt Mitigation in Multilevel Converters

学校读者我要写书评

暂无评论

FCS-MPC with Computational Efficiency and dv/dt Mitigation i...

2024 IEEE Energy Conversion Congress and Exposition, ECCE 2024

作者： Castillo, Cristian García, Cristian Sanchez, Daniel Araya, Baldomero Ma, Chenwei Davari, S. Alireza Long, Bo Rodriguez, Jose Universidad de Talca Dept. of Electrical Engineering Curico Chile Southwest Jiaotong University School of Electrical Engineering Chengdu China Universidad Andres Bello School of Electrical and Data Engineering Santiago Chile University of Electronic Science and Technology of China Institute of Electronic and Information Engineering Guangdong China Universidad San Sebastian Center for Energy Transition Santiago Chile

ISBN: (纸本)9798350376067

Finite Control Set Model Predictive Control (FCS-MPC) provides accurate voltage and current control in multilevel converters, but computational complexity remains, especially in medium voltage (MV) and high power (HP) applications. This paper proposes an enhanced control strategy for a 5-level Active NPC converter, improving the approach presented in [1] to reduce computational burden and mitigate undesired voltage level jumps and high dv/dt. Simulation results demonstrate improved performance compared to standard FCS-MPC and the original [1] strategy, showcasing reduced voltage level jumps and enhanced current tracking in steady-state and transient scenarios. The proposed enhancement shows promise in improving the efficiency and reliability of multilevel converters in MV and HP applications, with implications for further research in control optimization and performance enhancement. © 2024 IEEE.

关键词： Model predictive control

Approximation of RKHS Functionals by Neural Networks

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhou, Tian-Yi Suh, Namjoon Cheng, Guang Huo, Xiaoming H. Milton Stewart School of Industrial and Systems Engineering Georgia Institute of Technology United States Department of Statistics & Data Science UCLA United States

Motivated by the abundance of functional data such as time series and images, there has been a growing interest in integrating such data into neural networks and learning maps from function spaces to R (i.e., functionals). In this paper, we study the approximation of functionals on reproducing kernel Hilbert spaces (RKHS’s) using neural networks. We establish the universality of the approximation of functionals on the RKHS’s. Specifically, we derive explicit error bounds for those induced by inverse multiquadric, Gaussian, and Sobolev kernels. Moreover, we apply our findings to functional regression, proving that neural networks can accurately approximate the regression maps in generalized functional linear models. Existing works on functional learning require integration-type basis function expansions with a set of pre-specified basis functions. By leveraging the interpolating orthogonal projections in RKHS’s, our proposed network is much simpler in that we use point evaluations to replace basis function expansions. © 2024, CC BY.

关键词： Vector spaces