Pub Date : 2017-06-16DOI: 10.1109/ICASSP.2017.7953241
Kuan-Yu Chen, Shih-Hung Liu, Berlin Chen, H. Wang
Because unprecedented volumes of multimedia data associated with spoken documents have been made available to the public, spoken document retrieval (SDR) has become an important research area in the past decades. Recently, representation learning has emerged as an active research topic in many machine learning applications owing largely to its excellent performance. In the context of natural language processing, the pioneering work can date back to the word embedding methods. However, learning of paragraph (or sentence and document) representations is more reasonable and suitable for some tasks, such as information retrieval and document summarization. Nevertheless, as far as we are aware, there is relatively less work focusing on launching paragraph embedding methods into SDR. Motivated by these observations, this paper proposes a novel paragraph embedding method, named the locality-preserving essence vector (LPEV) model. LPEV is designed with consideration to two aspects. First, the model aims at not only distilling the most representative information from a paragraph but also getting rid of the general background information. Second, inspired by the local invariance perspective, which is a celebrated principle used in manifold learning techniques, LPEV also manages to preserve semantic locality in the learned low-dimensional embedding space for producing more informative and discriminative vector representations of paragraphs. On top of the proposed framework, a series of empirical SDR experiments conducted on the TDT-2 (Topic Detection and Tracking) collection demonstrate the good efficacy of our SDR methods as compared to existing strong baselines.
由于与口语文件相关的多媒体数据量空前，在过去的几十年里，口语文件检索(SDR)已成为一个重要的研究领域。近年来，表征学习因其优异的性能在许多机器学习应用中成为一个活跃的研究课题。在自然语言处理的背景下，开创性的工作可以追溯到词嵌入方法。然而，段落(或句子和文档)表征的学习更合理，更适合一些任务，如信息检索和文档摘要。然而，据我们所知，将段落嵌入方法引入SDR的工作相对较少。基于这些观察结果，本文提出了一种新的段落嵌入方法，称为位置保持本质向量(LPEV)模型。LPEV的设计考虑了两个方面。首先，该模型旨在从段落中提炼出最具代表性的信息，同时去除一般背景信息。其次，受局部不变性视角(这是流形学习技术中使用的著名原理)的启发，LPEV还设法在学习的低维嵌入空间中保持语义局部性，从而产生更有信息和判别性的段落向量表示。在提出的框架之上，在TDT-2 (Topic Detection and Tracking)数据集上进行的一系列SDR实证实验表明，与现有的强基线相比，我们的SDR方法具有良好的有效性。
{"title":"A locality-preserving essence vector modeling framework for spoken document retrieval","authors":"Kuan-Yu Chen, Shih-Hung Liu, Berlin Chen, H. Wang","doi":"10.1109/ICASSP.2017.7953241","DOIUrl":"https://doi.org/10.1109/ICASSP.2017.7953241","url":null,"abstract":"Because unprecedented volumes of multimedia data associated with spoken documents have been made available to the public, spoken document retrieval (SDR) has become an important research area in the past decades. Recently, representation learning has emerged as an active research topic in many machine learning applications owing largely to its excellent performance. In the context of natural language processing, the pioneering work can date back to the word embedding methods. However, learning of paragraph (or sentence and document) representations is more reasonable and suitable for some tasks, such as information retrieval and document summarization. Nevertheless, as far as we are aware, there is relatively less work focusing on launching paragraph embedding methods into SDR. Motivated by these observations, this paper proposes a novel paragraph embedding method, named the locality-preserving essence vector (LPEV) model. LPEV is designed with consideration to two aspects. First, the model aims at not only distilling the most representative information from a paragraph but also getting rid of the general background information. Second, inspired by the local invariance perspective, which is a celebrated principle used in manifold learning techniques, LPEV also manages to preserve semantic locality in the learned low-dimensional embedding space for producing more informative and discriminative vector representations of paragraphs. On top of the proposed framework, a series of empirical SDR experiments conducted on the TDT-2 (Topic Detection and Tracking) collection demonstrate the good efficacy of our SDR methods as compared to existing strong baselines.","PeriodicalId":118243,"journal":{"name":"2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","volume":"85 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121250510","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-16DOI: 10.1109/ICASSP.2017.7953053
A. Chiş, J. Lundén, V. Koivunen
In this paper we propose a novel coalitional game theory based optimization method for minimizing the cost of the electricity consumed from the power grid by a community of smart households. A smart household may own both a renewable energy source and an energy storage system (ESS), or only an ESS. We propose an optimization model in which all the members of the community jointly share their renewable resources and storage systems. We show that the proposed coalitional optimization method reduces the consumption costs both at community level and at the individual level when compared to the case in which the households would individually optimize their costs. The monetary revenues gained by the coalition are divided among the members of the coalition according to the Shapley value. Simulation examples show that the proposed coalitional optimization method may reduce the electricity costs for the community by roughly 18%.
{"title":"Coalitional game theoretic optimization of electricity cost for communities of smart households","authors":"A. Chiş, J. Lundén, V. Koivunen","doi":"10.1109/ICASSP.2017.7953053","DOIUrl":"https://doi.org/10.1109/ICASSP.2017.7953053","url":null,"abstract":"In this paper we propose a novel coalitional game theory based optimization method for minimizing the cost of the electricity consumed from the power grid by a community of smart households. A smart household may own both a renewable energy source and an energy storage system (ESS), or only an ESS. We propose an optimization model in which all the members of the community jointly share their renewable resources and storage systems. We show that the proposed coalitional optimization method reduces the consumption costs both at community level and at the individual level when compared to the case in which the households would individually optimize their costs. The monetary revenues gained by the coalition are divided among the members of the coalition according to the Shapley value. Simulation examples show that the proposed coalitional optimization method may reduce the electricity costs for the community by roughly 18%.","PeriodicalId":118243,"journal":{"name":"2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","volume":"161 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121503668","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-16DOI: 10.1109/ICASSP.2017.7953059
Puyang Wang, Vishal M. Patel
Fourier descriptors (FDs) are shape-based features for the recognition of two-dimensional connected shapes. We propose a method that can extract FDs of an object directly from compressive measurements without reconstructing the image. Our method entails estimating the edges via discrete horizontal and vertical image gradients from compressive measurements. Fourier descriptors are then extracted from the thresholded edges. One of the main advantages of the proposed method is that it requires fewer number of compressive measurements to estimate FDs than required to estimate the original image. Various numerical experiments on synthetic and real data demonstrate the effectiveness of the proposed method.
{"title":"Extracting Fourier descriptors from compressive measurements","authors":"Puyang Wang, Vishal M. Patel","doi":"10.1109/ICASSP.2017.7953059","DOIUrl":"https://doi.org/10.1109/ICASSP.2017.7953059","url":null,"abstract":"Fourier descriptors (FDs) are shape-based features for the recognition of two-dimensional connected shapes. We propose a method that can extract FDs of an object directly from compressive measurements without reconstructing the image. Our method entails estimating the edges via discrete horizontal and vertical image gradients from compressive measurements. Fourier descriptors are then extracted from the thresholded edges. One of the main advantages of the proposed method is that it requires fewer number of compressive measurements to estimate FDs than required to estimate the original image. Various numerical experiments on synthetic and real data demonstrate the effectiveness of the proposed method.","PeriodicalId":118243,"journal":{"name":"2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126057638","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-16DOI: 10.1109/ICASSP.2017.7953291
Gita Babazadeh Eslamlou, A. Jung, N. Goertz
We consider the problem of recovering a smooth graph signal from noisy samples observed at a small number of nodes. The signal recovery is formulated as a convex optimization problem using Tikhonov regularization based on the graph Laplacian quadratic form. The optimality conditions for this optimization problem form a system of linear equations involving the graph Laplacian. We solve this linear system via the iterative Gauss-Seidel method, which is shown to be particularly well-suited for smooth graph signal recovery. The effectiveness of the proposed recovery method is verified by numerical experiments using a real-world data-set.
{"title":"Smooth graph signal recovery via efficient Laplacian solvers","authors":"Gita Babazadeh Eslamlou, A. Jung, N. Goertz","doi":"10.1109/ICASSP.2017.7953291","DOIUrl":"https://doi.org/10.1109/ICASSP.2017.7953291","url":null,"abstract":"We consider the problem of recovering a smooth graph signal from noisy samples observed at a small number of nodes. The signal recovery is formulated as a convex optimization problem using Tikhonov regularization based on the graph Laplacian quadratic form. The optimality conditions for this optimization problem form a system of linear equations involving the graph Laplacian. We solve this linear system via the iterative Gauss-Seidel method, which is shown to be particularly well-suited for smooth graph signal recovery. The effectiveness of the proposed recovery method is verified by numerical experiments using a real-world data-set.","PeriodicalId":118243,"journal":{"name":"2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134619066","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-16DOI: 10.1109/ICASSP.2017.7952411
Takanori Fujisawa, M. Ikehara
This paper improves a colorization-based image coding using image segmentation and adaptive colorspaces. Recently, various approaches for color image coding based on colorization have been presented. These methods utilize a YCbCr colorspace and transfer the luminance component by a conventional compression method. Then, the chrominance components are approximated from the luminance component using a colorization method. Our method segments a luminance component into small segments called superpixels, and reconstructs the chrominance of each superpixel as a linear combination of its luminance. For chrominance components, we introduce an adaptive color space transform optimized for liner combination. This is because YCbCr colorspace cannot always become a good approximation of the chrominance. In addition, we introduce an automatic selection for the number of superpixel segments from a given quality factor. The simulation with standard images shows that our method performs better result than conventional coding schemes.
{"title":"Color image coding based on linear combination of adaptive colorspaces","authors":"Takanori Fujisawa, M. Ikehara","doi":"10.1109/ICASSP.2017.7952411","DOIUrl":"https://doi.org/10.1109/ICASSP.2017.7952411","url":null,"abstract":"This paper improves a colorization-based image coding using image segmentation and adaptive colorspaces. Recently, various approaches for color image coding based on colorization have been presented. These methods utilize a YCbCr colorspace and transfer the luminance component by a conventional compression method. Then, the chrominance components are approximated from the luminance component using a colorization method. Our method segments a luminance component into small segments called superpixels, and reconstructs the chrominance of each superpixel as a linear combination of its luminance. For chrominance components, we introduce an adaptive color space transform optimized for liner combination. This is because YCbCr colorspace cannot always become a good approximation of the chrominance. In addition, we introduce an automatic selection for the number of superpixel segments from a given quality factor. The simulation with standard images shows that our method performs better result than conventional coding schemes.","PeriodicalId":118243,"journal":{"name":"2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128875247","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-16DOI: 10.1109/ICASSP.2017.7953340
Pascal Zille, V. Calhoun, J. Stephen, T. Wilson, Yu-ping Wang
Functional magnetic resonance imaging (fMRI) is a powerful tool to analyze brain development and neuronal activity. Identifying discriminative brain regions between various groups within a population has generated great interest in recent years. In this work, we consider the problem of estimating multiple sparse, co-activated brain regions from fMRI observations belonging to different classes. More precisely, we propose a method to analyze functional connectivity differences between children and young adults. Often, analysis is conducted on each class separately. Here, we propose to rely on a generalized fused Lasso penalty to extract both class-specific and shared co-expressed regions. In order to validate our method, experiments are performed on an fMRI dataset comprised of normally developing children from 8 to 21. The results demonstrate that the proposed method is able to properly extract meaningful sub-networks, which results in improved classification accuracy between the two classes.
{"title":"Fused estimation of sparse connectivity patterns from rest fMRI","authors":"Pascal Zille, V. Calhoun, J. Stephen, T. Wilson, Yu-ping Wang","doi":"10.1109/ICASSP.2017.7953340","DOIUrl":"https://doi.org/10.1109/ICASSP.2017.7953340","url":null,"abstract":"Functional magnetic resonance imaging (fMRI) is a powerful tool to analyze brain development and neuronal activity. Identifying discriminative brain regions between various groups within a population has generated great interest in recent years. In this work, we consider the problem of estimating multiple sparse, co-activated brain regions from fMRI observations belonging to different classes. More precisely, we propose a method to analyze functional connectivity differences between children and young adults. Often, analysis is conducted on each class separately. Here, we propose to rely on a generalized fused Lasso penalty to extract both class-specific and shared co-expressed regions. In order to validate our method, experiments are performed on an fMRI dataset comprised of normally developing children from 8 to 21. The results demonstrate that the proposed method is able to properly extract meaningful sub-networks, which results in improved classification accuracy between the two classes.","PeriodicalId":118243,"journal":{"name":"2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122349428","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-16DOI: 10.1109/ICASSP.2017.7953209
B. Bollepalli, Manu Airaksinen, P. Alku
In statistical parametric speech synthesis (SPSS), a few studies have investigated the Lombard effect, specifically by using hidden Markov model (HMM)-based systems. Recently, artificial neural networks have demonstrated promising results in SPSS, specifically by using long short-term memory recurrent neural networks (LSTMs). The Lombard effect, however, has not been studied in the LSTM-based speech synthesis systems. In this study, we propose three methods for Lombard speech adaptation in LSTM-based speech synthesis. In particular, (1) we augment Lombard specific information with the linguistic features as input, (2) scale the hidden activations using the learning hidden unit contributions (LHUC) method, and (3) fine-tune the LSTMs trained on normal speech with a small Lombard speech data. To investigate the effectiveness of the proposed methods, we carry out experiments using small (10 utterances) and large (500 utterances) Lombard speech data. Experimental results confirm the adaptability of the LSTMs, and similarity tests show that the LSTMs can achieve significantly better adaptation performance than the HMMs in both small and large data conditions.
{"title":"Lombard speech synthesis using long short-term memory recurrent neural networks","authors":"B. Bollepalli, Manu Airaksinen, P. Alku","doi":"10.1109/ICASSP.2017.7953209","DOIUrl":"https://doi.org/10.1109/ICASSP.2017.7953209","url":null,"abstract":"In statistical parametric speech synthesis (SPSS), a few studies have investigated the Lombard effect, specifically by using hidden Markov model (HMM)-based systems. Recently, artificial neural networks have demonstrated promising results in SPSS, specifically by using long short-term memory recurrent neural networks (LSTMs). The Lombard effect, however, has not been studied in the LSTM-based speech synthesis systems. In this study, we propose three methods for Lombard speech adaptation in LSTM-based speech synthesis. In particular, (1) we augment Lombard specific information with the linguistic features as input, (2) scale the hidden activations using the learning hidden unit contributions (LHUC) method, and (3) fine-tune the LSTMs trained on normal speech with a small Lombard speech data. To investigate the effectiveness of the proposed methods, we carry out experiments using small (10 utterances) and large (500 utterances) Lombard speech data. Experimental results confirm the adaptability of the LSTMs, and similarity tests show that the LSTMs can achieve significantly better adaptation performance than the HMMs in both small and large data conditions.","PeriodicalId":118243,"journal":{"name":"2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121688679","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-16DOI: 10.1109/ICASSP.2017.7953421
F. Gregorio, Gustavo J. González, J. Cousseau, T. Riihonen, R. Wichman
Genuine full-duplex operation requires effective mitigation of self-interference (SI) due to simultaneous transmission and reception at the same frequency band. In addition to its well-known harmful effect on signals of high peak-to-average power ratio, nonlinear behavior of a power amplifier (PA) complicates SI cancellation and induces spectral regrowth. We introduce a digital predistortion architecture that linearizes the response of the PA in a full-duplex transceiver. Specifically, we propose a two-step procedure to estimate the predistorter parameters and the SI canceller coefficients. Different from a direct application of conventional predistorters, the proposed architecture does not need an extra RF chain to estimate the PA response and exploits the inherent SI signal instead. Finally, simulation results show that the proposed scheme is able to increase significantly the signal-to-interference-plus-noise ratio at the transceiver output and to reduce out-of-band emissions when compared to linear and nonlinear cancellation without predistortion.
{"title":"Predistortion for power amplifier linearization in full-duplex transceivers without extra RF chain","authors":"F. Gregorio, Gustavo J. González, J. Cousseau, T. Riihonen, R. Wichman","doi":"10.1109/ICASSP.2017.7953421","DOIUrl":"https://doi.org/10.1109/ICASSP.2017.7953421","url":null,"abstract":"Genuine full-duplex operation requires effective mitigation of self-interference (SI) due to simultaneous transmission and reception at the same frequency band. In addition to its well-known harmful effect on signals of high peak-to-average power ratio, nonlinear behavior of a power amplifier (PA) complicates SI cancellation and induces spectral regrowth. We introduce a digital predistortion architecture that linearizes the response of the PA in a full-duplex transceiver. Specifically, we propose a two-step procedure to estimate the predistorter parameters and the SI canceller coefficients. Different from a direct application of conventional predistorters, the proposed architecture does not need an extra RF chain to estimate the PA response and exploits the inherent SI signal instead. Finally, simulation results show that the proposed scheme is able to increase significantly the signal-to-interference-plus-noise ratio at the transceiver output and to reduce out-of-band emissions when compared to linear and nonlinear cancellation without predistortion.","PeriodicalId":118243,"journal":{"name":"2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115344345","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-16DOI: 10.1109/ICASSP.2017.7952777
Tommaso Piatti, Shiwen Lei, J. Barras, A. Jakobsson
Given its high specificity, the use of nuclear quadrupole resonance (NQR) spectroscopy allows for a reliable identification and quantification of substances containing quadrupolar nuclei, such as the 14N nucleus prevalent in many explosives, medicines, and narcotics. Regrettably, the measured signals are typically weak and suffers from interference signals often being several orders of magnitude stronger than the signal of interest. In this work, we propose a two-channel setup allowing for interference cancellation in applications such as demining. The proposed techniques forms an estimate of the interference using the secondary channel, and then removes it from the primary channel. The improved performance of the resulting detector is illustrated using real measurements of NaNO2.
{"title":"Interference cancellation in two-channel nuclear quadrupole resonance measurements","authors":"Tommaso Piatti, Shiwen Lei, J. Barras, A. Jakobsson","doi":"10.1109/ICASSP.2017.7952777","DOIUrl":"https://doi.org/10.1109/ICASSP.2017.7952777","url":null,"abstract":"Given its high specificity, the use of nuclear quadrupole resonance (NQR) spectroscopy allows for a reliable identification and quantification of substances containing quadrupolar nuclei, such as the 14N nucleus prevalent in many explosives, medicines, and narcotics. Regrettably, the measured signals are typically weak and suffers from interference signals often being several orders of magnitude stronger than the signal of interest. In this work, we propose a two-channel setup allowing for interference cancellation in applications such as demining. The proposed techniques forms an estimate of the interference using the secondary channel, and then removes it from the primary channel. The improved performance of the resulting detector is illustrated using real measurements of NaNO2.","PeriodicalId":118243,"journal":{"name":"2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130788518","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2017-06-16DOI: 10.1109/ICASSP.2017.7952266
Chen Tang, Norihito Inamuro, Takashi Ijiri, A. Hirabayashi
The compressed sensing using dictionary learning has led to state-of-the-art results for magnetic resonance imaging (MRI) reconstruction from highly under-sampled measurements. Dictionary learning had been considered time-consuming especially when the patch size or the number of training patches is large. Recently, double sparsity model and online dictionary learning algorithm were proposed to obtain dictionaries with much less computational time. In this paper, we propose an efficient MRI reconstruction method by adopting the double sparsity model with the online dictionary learning method. Besides, for better reconstruction, we use separately prepared fully-sampled MRI images to train dictionaries. We compare results of the proposed technique to traditional offline methods with and without double sparsity model. Our simulation results show that the proposed technique is approximately twice faster than the traditional methods while maintaining the same reconstruction quality. Furthermore, our technique performed even better for lower sampling rate.
{"title":"Compressed sensing MRI using double sparsity with additional training images","authors":"Chen Tang, Norihito Inamuro, Takashi Ijiri, A. Hirabayashi","doi":"10.1109/ICASSP.2017.7952266","DOIUrl":"https://doi.org/10.1109/ICASSP.2017.7952266","url":null,"abstract":"The compressed sensing using dictionary learning has led to state-of-the-art results for magnetic resonance imaging (MRI) reconstruction from highly under-sampled measurements. Dictionary learning had been considered time-consuming especially when the patch size or the number of training patches is large. Recently, double sparsity model and online dictionary learning algorithm were proposed to obtain dictionaries with much less computational time. In this paper, we propose an efficient MRI reconstruction method by adopting the double sparsity model with the online dictionary learning method. Besides, for better reconstruction, we use separately prepared fully-sampled MRI images to train dictionaries. We compare results of the proposed technique to traditional offline methods with and without double sparsity model. Our simulation results show that the proposed technique is approximately twice faster than the traditional methods while maintaining the same reconstruction quality. Furthermore, our technique performed even better for lower sampling rate.","PeriodicalId":118243,"journal":{"name":"2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123837853","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：