Anomaly detection (AD) is increasingly recognized as a key component for ensuring the resilience of future communication systems. While deep learning has shown state-of-the-art AD performance, its application in critical systems is hindered by concerns regarding training data efficiency, domain adaptation and interpretability. This work considers AD in network flows using incomplete measurements, leveraging a robust tensor decomposition approach and deep unrolling techniques to address these challenges. We first propose a novel block-successive convex approximation algorithm based on a regularized model-fitting objective where the normal flows are modeled as low-rank tensors and anomalies as sparse. An augmentation of the objective is introduced to decrease the computational cost. We apply deep unrolling to derive a novel deep network architecture based on our proposed algorithm, treating the regularization parameters as learnable weights. Inspired by Bayesian approaches, we extend the model architecture to perform online adaptation to per-flow and per-time-step statistics, improving AD performance while maintaining a low parameter count and preserving the problem's permutation equivariances. To optimize the deep network weights for detection performance, we employ a homotopy optimization approach based on an efficient approximation of the area under the receiver operating characteristic curve. Extensive experiments on synthetic and real-world data demonstrate that our proposed deep network architecture exhibits a high training data efficiency, outperforms reference methods, and adapts seamlessly to varying network topologies.
异常检测（AD）越来越被认为是确保未来通信系统弹性的关键组成部分。虽然深度学习已经显示出最先进的异常检测性能，但其在关键系统中的应用却受到训练数据效率、领域适应性和可解释性等问题的阻碍。本研究利用不完整的测量数据考虑网络流中的反向增量，并利用稳健的张量分解方法和深度滚动技术来应对这些挑战。我们首先提出了一种基于正则化模型拟合目标的新型块继承凸近似算法，其中正常流量被建模为低秩张量，异常流量被建模为稀疏。为了降低计算成本，我们引入了一个增强目标。我们基于所提出的算法，应用深度开卷法推导出一种新型的深度网络架构，并将其标准化参数视为可学习的权重。在贝叶斯方法的启发下，我们扩展了模型架构，以对每流和每时间步统计进行在线适应，从而提高了 AD 性能，同时保持了较低的参数数量，并保留了问题的包换方差。为了优化深度网络权重以提高检测性能，我们采用了一种同调优化方法，该方法基于对接收器工作特征曲线下面积的有效近似。在合成数据和真实世界数据上进行的广泛实验表明，我们提出的深度网络架构具有很高的训练数据效率，优于参考方法，并能无缝适应不同的网络拓扑结构。
{"title":"Adaptive Anomaly Detection in Network Flows with Low-Rank Tensor Decompositions and Deep Unrolling","authors":"Lukas Schynol, Marius Pesavento","doi":"arxiv-2409.11529","DOIUrl":"https://doi.org/arxiv-2409.11529","url":null,"abstract":"Anomaly detection (AD) is increasingly recognized as a key component for\u0000ensuring the resilience of future communication systems. While deep learning\u0000has shown state-of-the-art AD performance, its application in critical systems\u0000is hindered by concerns regarding training data efficiency, domain adaptation\u0000and interpretability. This work considers AD in network flows using incomplete\u0000measurements, leveraging a robust tensor decomposition approach and deep\u0000unrolling techniques to address these challenges. We first propose a novel\u0000block-successive convex approximation algorithm based on a regularized\u0000model-fitting objective where the normal flows are modeled as low-rank tensors\u0000and anomalies as sparse. An augmentation of the objective is introduced to\u0000decrease the computational cost. We apply deep unrolling to derive a novel deep\u0000network architecture based on our proposed algorithm, treating the\u0000regularization parameters as learnable weights. Inspired by Bayesian\u0000approaches, we extend the model architecture to perform online adaptation to\u0000per-flow and per-time-step statistics, improving AD performance while\u0000maintaining a low parameter count and preserving the problem's permutation\u0000equivariances. To optimize the deep network weights for detection performance,\u0000we employ a homotopy optimization approach based on an efficient approximation\u0000of the area under the receiver operating characteristic curve. Extensive\u0000experiments on synthetic and real-world data demonstrate that our proposed deep\u0000network architecture exhibits a high training data efficiency, outperforms\u0000reference methods, and adapts seamlessly to varying network topologies.","PeriodicalId":501034,"journal":{"name":"arXiv - EE - Signal Processing","volume":"4 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142251314","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}
Gerard I. Gállego, Roy Fejgin, Chunghsin Yeh, Xiaoyu Liu, Gautam Bhattacharya
Audio token modeling has become a powerful framework for speech synthesis, with two-stage approaches employing semantic tokens remaining prevalent. In this paper, we aim to simplify this process by introducing a semantic knowledge distillation method that enables high-quality speech generation in a single stage. Our proposed model improves speech quality, intelligibility, and speaker similarity compared to a single-stage baseline. Although two-stage systems still lead in intelligibility, our model significantly narrows the gap while delivering comparable speech quality. These findings showcase the potential of single-stage models to achieve efficient, high-quality TTS with a more compact and streamlined architecture.
{"title":"Single-stage TTS with Masked Audio Token Modeling and Semantic Knowledge Distillation","authors":"Gerard I. Gállego, Roy Fejgin, Chunghsin Yeh, Xiaoyu Liu, Gautam Bhattacharya","doi":"arxiv-2409.11003","DOIUrl":"https://doi.org/arxiv-2409.11003","url":null,"abstract":"Audio token modeling has become a powerful framework for speech synthesis,\u0000with two-stage approaches employing semantic tokens remaining prevalent. In\u0000this paper, we aim to simplify this process by introducing a semantic knowledge\u0000distillation method that enables high-quality speech generation in a single\u0000stage. Our proposed model improves speech quality, intelligibility, and speaker\u0000similarity compared to a single-stage baseline. Although two-stage systems\u0000still lead in intelligibility, our model significantly narrows the gap while\u0000delivering comparable speech quality. These findings showcase the potential of\u0000single-stage models to achieve efficient, high-quality TTS with a more compact\u0000and streamlined architecture.","PeriodicalId":501034,"journal":{"name":"arXiv - EE - Signal Processing","volume":"38 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142251323","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}
Zachary Schutz, Daniel J. Jakubisin, Charles E. Thornton, R. Michael Buehrer
We study jamming of an OFDM-modulated signal which employs forward error correction coding. We extend this to leverage reinforcement learning with a contextual bandit to jam a 5G-based system implementing some aspects of the 5G protocol. This model introduces unreliable reward feedback in the form of ACK/NACK observations to the jammer to understand the effect of how imperfect observations of errors can affect the jammer's ability to learn. We gain insights into the convergence time of the jammer and its ability to jam a victim 5G waveform, as well as insights into the vulnerabilities of wireless communications for reinforcement learning-based jamming.
{"title":"Linear Jamming Bandits: Learning to Jam 5G-based Coded Communications Systems","authors":"Zachary Schutz, Daniel J. Jakubisin, Charles E. Thornton, R. Michael Buehrer","doi":"arxiv-2409.11191","DOIUrl":"https://doi.org/arxiv-2409.11191","url":null,"abstract":"We study jamming of an OFDM-modulated signal which employs forward error\u0000correction coding. We extend this to leverage reinforcement learning with a\u0000contextual bandit to jam a 5G-based system implementing some aspects of the 5G\u0000protocol. This model introduces unreliable reward feedback in the form of\u0000ACK/NACK observations to the jammer to understand the effect of how imperfect\u0000observations of errors can affect the jammer's ability to learn. We gain\u0000insights into the convergence time of the jammer and its ability to jam a\u0000victim 5G waveform, as well as insights into the vulnerabilities of wireless\u0000communications for reinforcement learning-based jamming.","PeriodicalId":501034,"journal":{"name":"arXiv - EE - Signal Processing","volume":"6 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142251317","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}
We address the Normalized Signal to Noise Ratio (NSNR) metric defined in the seminal paper by Reed, Mallett and Brennan on adaptive detection. The setting is detection of a target vector in additive correlated noise. NSNR is the ratio between the SNR of a linear detector which uses an estimated noise covariance and the SNR of clairvoyant detector based on the exact unknown covariance. It is not obvious how to evaluate NSNR since it is a function of the target vector. To close this gap, we consider the NSNR associated with the worst target. Using the Kantorovich Inequality, we provide a closed form solution for the worst case NSNR. Then, we prove that the classical Gaussian Kullback Leibler (KL) divergence bounds it. Numerical experiments with different true covariances and various estimates also suggest that the KL metric is more correlated with the NSNR metric than competing norm based metrics.
{"title":"On the normalized signal to noise ratio in covariance estimation","authors":"Tzvi Diskin, Ami Wiesel","doi":"arxiv-2409.10896","DOIUrl":"https://doi.org/arxiv-2409.10896","url":null,"abstract":"We address the Normalized Signal to Noise Ratio (NSNR) metric defined in the\u0000seminal paper by Reed, Mallett and Brennan on adaptive detection. The setting\u0000is detection of a target vector in additive correlated noise. NSNR is the ratio\u0000between the SNR of a linear detector which uses an estimated noise covariance\u0000and the SNR of clairvoyant detector based on the exact unknown covariance. It\u0000is not obvious how to evaluate NSNR since it is a function of the target\u0000vector. To close this gap, we consider the NSNR associated with the worst\u0000target. Using the Kantorovich Inequality, we provide a closed form solution for\u0000the worst case NSNR. Then, we prove that the classical Gaussian Kullback\u0000Leibler (KL) divergence bounds it. Numerical experiments with different true\u0000covariances and various estimates also suggest that the KL metric is more\u0000correlated with the NSNR metric than competing norm based metrics.","PeriodicalId":501034,"journal":{"name":"arXiv - EE - Signal Processing","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142251320","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}
The extended near-field range in future mm-Wave and sub-THz wireless networks demands a precise and efficient near-field channel simulator for understanding and optimizing wireless communications in this less-explored regime. This paper presents NirvaWave, a novel near-field channel simulator, built on scalar diffraction theory and Fourier principles, to provide precise wave propagation response in complex wireless mediums under custom user-defined transmitted EM signals. NirvaWave offers an interface for investigating novel near-field wavefronts, e.g., Airy beams, Bessel beams, and the interaction of mmWave and sub-THz signals with obstructions, reflectors, and scatterers. The simulation run-time in NirvaWave is orders of magnitude lower than its EM software counterparts that directly solve Maxwell Equations. Hence, NirvaWave enables a user-friendly interface for large-scale channel simulations required for developing new model-driven and data-driven techniques. We evaluated the performance of NirvaWave through direct comparison with EM simulation software. Finally, we have open-sourced the core codebase of NirvaWave in our GitHub repository (https://github.com/vahidyazdnian1378/NirvaWave).
{"title":"NirvaWave: An Accurate and Efficient Near Field Wave Propagation Simulator for 6G and Beyond","authors":"Vahid Yazdnian, Yasaman Ghasempour","doi":"arxiv-2409.11293","DOIUrl":"https://doi.org/arxiv-2409.11293","url":null,"abstract":"The extended near-field range in future mm-Wave and sub-THz wireless networks\u0000demands a precise and efficient near-field channel simulator for understanding\u0000and optimizing wireless communications in this less-explored regime. This paper\u0000presents NirvaWave, a novel near-field channel simulator, built on scalar\u0000diffraction theory and Fourier principles, to provide precise wave propagation\u0000response in complex wireless mediums under custom user-defined transmitted EM\u0000signals. NirvaWave offers an interface for investigating novel near-field\u0000wavefronts, e.g., Airy beams, Bessel beams, and the interaction of mmWave and\u0000sub-THz signals with obstructions, reflectors, and scatterers. The simulation\u0000run-time in NirvaWave is orders of magnitude lower than its EM software\u0000counterparts that directly solve Maxwell Equations. Hence, NirvaWave enables a\u0000user-friendly interface for large-scale channel simulations required for\u0000developing new model-driven and data-driven techniques. We evaluated the\u0000performance of NirvaWave through direct comparison with EM simulation software.\u0000Finally, we have open-sourced the core codebase of NirvaWave in our GitHub\u0000repository (https://github.com/vahidyazdnian1378/NirvaWave).","PeriodicalId":501034,"journal":{"name":"arXiv - EE - Signal Processing","volume":"28 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142251318","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}
Photoacoustic computed tomography (PACT) is a non-invasive imaging modality with wide medical applications. Conventional PACT image reconstruction algorithms suffer from wavefront distortion caused by the heterogeneous speed of sound (SOS) in tissue, which leads to image degradation. Accounting for these effects improves image quality, but measuring the SOS distribution is experimentally expensive. An alternative approach is to perform joint reconstruction of the initial pressure image and SOS using only the PA signals. Existing joint reconstruction methods come with limitations: high computational cost, inability to directly recover SOS, and reliance on inaccurate simplifying assumptions. Implicit neural representation, or neural fields, is an emerging technique in computer vision to learn an efficient and continuous representation of physical fields with a coordinate-based neural network. In this work, we introduce NF-APACT, an efficient self-supervised framework utilizing neural fields to estimate the SOS in service of an accurate and robust multi-channel deconvolution. Our method removes SOS aberrations an order of magnitude faster and more accurately than existing methods. We demonstrate the success of our method on a novel numerical phantom as well as an experimentally collected phantom and in vivo data. Our code and numerical phantom are available at https://github.com/Lukeli0425/NF-APACT.
光声计算机断层扫描（PACT）是一种无创成像模式，在医学上有着广泛的应用。传统的光声计算机断层扫描图像重建算法受到组织中异质声速（SOS）引起的波前失真影响，导致图像质量下降。考虑到这些影响可以提高图像质量，但测量 SOS 分布的实验成本很高。现有的联合重建方法有其局限性：计算成本高、无法直接恢复 SOS 以及依赖不准确的简化假设。隐式神经表示或神经场是计算机视觉领域的一种新兴技术，通过基于坐标的神经网络学习物理场的高效连续表示。在这项工作中，我们引入了 NF-APACT，这是一种高效的自我监督框架，利用神经场来估计 SOS，从而实现准确、稳健的多通道解卷积。与现有方法相比，我们的方法能更快更准确地消除 SOS 畸变。我们在一个新的数值模型以及实验收集的模型和体内数据上展示了我们方法的成功。我们的代码和数值模型可在 https://github.com/Lukeli0425/NF-APACT 网站上获得。
{"title":"Neural Fields for Adaptive Photoacoustic Computed Tomography","authors":"Tianao Li, Manxiu Cui, Cheng Ma, Emma Alexander","doi":"arxiv-2409.10876","DOIUrl":"https://doi.org/arxiv-2409.10876","url":null,"abstract":"Photoacoustic computed tomography (PACT) is a non-invasive imaging modality\u0000with wide medical applications. Conventional PACT image reconstruction\u0000algorithms suffer from wavefront distortion caused by the heterogeneous speed\u0000of sound (SOS) in tissue, which leads to image degradation. Accounting for\u0000these effects improves image quality, but measuring the SOS distribution is\u0000experimentally expensive. An alternative approach is to perform joint\u0000reconstruction of the initial pressure image and SOS using only the PA signals.\u0000Existing joint reconstruction methods come with limitations: high computational\u0000cost, inability to directly recover SOS, and reliance on inaccurate simplifying\u0000assumptions. Implicit neural representation, or neural fields, is an emerging\u0000technique in computer vision to learn an efficient and continuous\u0000representation of physical fields with a coordinate-based neural network. In\u0000this work, we introduce NF-APACT, an efficient self-supervised framework\u0000utilizing neural fields to estimate the SOS in service of an accurate and\u0000robust multi-channel deconvolution. Our method removes SOS aberrations an order\u0000of magnitude faster and more accurately than existing methods. We demonstrate\u0000the success of our method on a novel numerical phantom as well as an\u0000experimentally collected phantom and in vivo data. Our code and numerical\u0000phantom are available at https://github.com/Lukeli0425/NF-APACT.","PeriodicalId":501034,"journal":{"name":"arXiv - EE - Signal Processing","volume":"28 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142251325","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}
We propose a minimal power white box adversarial attack for Deep Learning based Automatic Modulation Classification (AMC). The proposed attack uses the Golden Ratio Search (GRS) method to find powerful attacks with minimal power. We evaluate the efficacy of the proposed method by comparing it with existing adversarial attack approaches. Additionally, we test the robustness of the proposed attack against various state-of-the-art architectures, including defense mechanisms such as adversarial training, binarization, and ensemble methods. Experimental results demonstrate that the proposed attack is powerful, requires minimal power, and can be generated in less time, significantly challenging the resilience of current AMC methods.
{"title":"Golden Ratio Search: A Low-Power Adversarial Attack for Deep Learning based Modulation Classification","authors":"Deepsayan Sadhukhan, Nitin Priyadarshini Shankar, Sheetal Kalyani","doi":"arxiv-2409.11454","DOIUrl":"https://doi.org/arxiv-2409.11454","url":null,"abstract":"We propose a minimal power white box adversarial attack for Deep Learning\u0000based Automatic Modulation Classification (AMC). The proposed attack uses the\u0000Golden Ratio Search (GRS) method to find powerful attacks with minimal power.\u0000We evaluate the efficacy of the proposed method by comparing it with existing\u0000adversarial attack approaches. Additionally, we test the robustness of the\u0000proposed attack against various state-of-the-art architectures, including\u0000defense mechanisms such as adversarial training, binarization, and ensemble\u0000methods. Experimental results demonstrate that the proposed attack is powerful,\u0000requires minimal power, and can be generated in less time, significantly\u0000challenging the resilience of current AMC methods.","PeriodicalId":501034,"journal":{"name":"arXiv - EE - Signal Processing","volume":"9 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142251312","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}
This paper presents an estimation method for time-varying graph signals among multiple sub-networks. In many sensor networks, signals observed are associated with nodes (i.e., sensors), and edges of the network represent the inter-node connectivity. For a large sensor network, measuring signal values at all nodes over time requires huge resources, particularly in terms of energy consumption. To alleviate the issue, we consider a scenario that a sub-network, i.e., cluster, from the whole network is extracted and an intra-cluster analysis is performed based on the statistics in the cluster. The statistics are then utilized to estimate signal values in another cluster. This leads to the requirement for transferring a set of parameters of the sub-network to the others, while the numbers of nodes in the clusters are typically different. In this paper, we propose a cooperative Kalman filter between two sub-networks. The proposed method alternately estimates signals in time between two sub-networks. We formulate a state-space model in the source cluster and transfer it to the target cluster on the basis of optimal transport. In the signal estimation experiments of synthetic and real-world signals, we validate the effectiveness of the proposed method.
{"title":"Time-Varying Graph Signal Estimation among Multiple Sub-Networks","authors":"Tsutahiro Fukuhara, Junya Hara, Hiroshi Higashi, Yuichi Tanaka","doi":"arxiv-2409.10915","DOIUrl":"https://doi.org/arxiv-2409.10915","url":null,"abstract":"This paper presents an estimation method for time-varying graph signals among\u0000multiple sub-networks. In many sensor networks, signals observed are associated\u0000with nodes (i.e., sensors), and edges of the network represent the inter-node\u0000connectivity. For a large sensor network, measuring signal values at all nodes\u0000over time requires huge resources, particularly in terms of energy consumption.\u0000To alleviate the issue, we consider a scenario that a sub-network, i.e.,\u0000cluster, from the whole network is extracted and an intra-cluster analysis is\u0000performed based on the statistics in the cluster. The statistics are then\u0000utilized to estimate signal values in another cluster. This leads to the\u0000requirement for transferring a set of parameters of the sub-network to the\u0000others, while the numbers of nodes in the clusters are typically different. In\u0000this paper, we propose a cooperative Kalman filter between two sub-networks.\u0000The proposed method alternately estimates signals in time between two\u0000sub-networks. We formulate a state-space model in the source cluster and\u0000transfer it to the target cluster on the basis of optimal transport. In the\u0000signal estimation experiments of synthetic and real-world signals, we validate\u0000the effectiveness of the proposed method.","PeriodicalId":501034,"journal":{"name":"arXiv - EE - Signal Processing","volume":"65 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142251321","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}
In reconfigurable intelligent surface (RIS) aided systems, the joint optimization of the precoder matrix at the base station and the phase shifts of the RIS elements involves significant complexity. In this paper, we propose a complex-valued, geometry aware meta-learning neural network that maximizes the weighted sum rate in a multi-user multiple input single output system. By leveraging the complex circle geometry for phase shifts and spherical geometry for the precoder, the optimization occurs on Riemannian manifolds, leading to faster convergence. We use a complex-valued neural network for phase shifts and an Euler inspired update for the precoder network. Our approach outperforms existing neural network-based algorithms, offering higher weighted sum rates, lower power consumption, and significantly faster convergence. Specifically, it converges faster by nearly 100 epochs, with a 0.7 bps improvement in weighted sum rate and a 1.8 dBm power gain when compared with existing work.
{"title":"Geometry Aware Meta-Learning Neural Network for Joint Phase and Precoder Optimization in RIS","authors":"Dahlia Devapriya, Sheetal Kalyani","doi":"arxiv-2409.11270","DOIUrl":"https://doi.org/arxiv-2409.11270","url":null,"abstract":"In reconfigurable intelligent surface (RIS) aided systems, the joint\u0000optimization of the precoder matrix at the base station and the phase shifts of\u0000the RIS elements involves significant complexity. In this paper, we propose a\u0000complex-valued, geometry aware meta-learning neural network that maximizes the\u0000weighted sum rate in a multi-user multiple input single output system. By\u0000leveraging the complex circle geometry for phase shifts and spherical geometry\u0000for the precoder, the optimization occurs on Riemannian manifolds, leading to\u0000faster convergence. We use a complex-valued neural network for phase shifts and\u0000an Euler inspired update for the precoder network. Our approach outperforms\u0000existing neural network-based algorithms, offering higher weighted sum rates,\u0000lower power consumption, and significantly faster convergence. Specifically, it\u0000converges faster by nearly 100 epochs, with a 0.7 bps improvement in weighted\u0000sum rate and a 1.8 dBm power gain when compared with existing work.","PeriodicalId":501034,"journal":{"name":"arXiv - EE - Signal Processing","volume":"54 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142251322","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}
Dorsa Ameri, Ali K. Z. Tehrani, Ivan M. Rosado-Mendez, Hassan Rivaz
Homodyned K-distribution (HK-distribution) parameter estimation in quantitative ultrasound (QUS) has been recently addressed using Bayesian Neural Networks (BNNs). BNNs have been shown to significantly reduce computational time in speckle statistics-based QUS without compromising accuracy and precision. Additionally, they provide estimates of feature uncertainty, which can guide the clinician's trust in the reported feature value. The total predictive uncertainty in Bayesian modeling can be decomposed into epistemic (uncertainty over the model parameters) and aleatoric (uncertainty inherent in the data) components. By decomposing the predictive uncertainty, we can gain insights into the factors contributing to the total uncertainty. In this study, we propose a method to compute epistemic and aleatoric uncertainties for HK-distribution parameters ($alpha$ and $k$) estimated by a BNN, in both simulation and experimental data. In addition, we investigate the relationship between the prediction error and both uncertainties, shedding light on the interplay between these uncertainties and HK parameters errors.
{"title":"Uncertainty Decomposition and Error Margin Detection of Homodyned-K Distribution in Quantitative Ultrasound","authors":"Dorsa Ameri, Ali K. Z. Tehrani, Ivan M. Rosado-Mendez, Hassan Rivaz","doi":"arxiv-2409.11583","DOIUrl":"https://doi.org/arxiv-2409.11583","url":null,"abstract":"Homodyned K-distribution (HK-distribution) parameter estimation in\u0000quantitative ultrasound (QUS) has been recently addressed using Bayesian Neural\u0000Networks (BNNs). BNNs have been shown to significantly reduce computational\u0000time in speckle statistics-based QUS without compromising accuracy and\u0000precision. Additionally, they provide estimates of feature uncertainty, which\u0000can guide the clinician's trust in the reported feature value. The total\u0000predictive uncertainty in Bayesian modeling can be decomposed into epistemic\u0000(uncertainty over the model parameters) and aleatoric (uncertainty inherent in\u0000the data) components. By decomposing the predictive uncertainty, we can gain\u0000insights into the factors contributing to the total uncertainty. In this study,\u0000we propose a method to compute epistemic and aleatoric uncertainties for\u0000HK-distribution parameters ($alpha$ and $k$) estimated by a BNN, in both\u0000simulation and experimental data. In addition, we investigate the relationship\u0000between the prediction error and both uncertainties, shedding light on the\u0000interplay between these uncertainties and HK parameters errors.","PeriodicalId":501034,"journal":{"name":"arXiv - EE - Signal Processing","volume":"65 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142251310","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：