检索结果-内蒙古大学图书馆

Dictionary Learning-Structured Reinforcement Learning With Adaptive-Sparsity Regularizer

IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS 2024年第2期60卷 1753-1769页

作者： Li, Zhenni Tang, Jianhao Zhao, Haoli Chen, Ci Xie, Shengli Guangdong Univ Technol Sch Automat Guangzhou 510006 Peoples R China Guangdong HongKong Macao Joint Lab Smart Discrete Guangzhou 510006 Guangdong Peoples R China 111 Ctr Intelligent Batch Mfg Based IoT Technol GD Guangzhou 510006 Peoples R China Guangdong Key Lab IoT Informat Technol GDUT Guangzhou 510006 Peoples R China Minist Educ Key Lab Intelligent Informat Proc & Syst Integrat Guangzhou 510006 Peoples R China

Deep reinforcement learning (DRL) has been applied to satellite navigation and positioning applications. Its performance relies heavily on the function-approximation capability of deep neural networks. However, existing DRL models suffer from catastrophic interference, resulting in inaccurate function approximation. The sparse-coding-based DRL is an effective method to mitigating this interference, but existing methods involve the following two challenging issues: first, the value function estimation network suffers from instability problems with gradient backpropagation, including gradient explosion and gradient vanishing, second, existing methods are limited to using hand-crafted sparse regularizers that produce only static sparsity, which may be difficult to apply in various dynamic reinforcement learning (RL) environments. In this article, we propose a novel dictionary learning (DL)-structured RL model with adaptive-sparsity regularizer (ASR) that alleviates the catastrophic interference and enables accurate value function approximation, thereby improving the RL performance. To alleviate the interference and avoid the instability problems in RL, a feedforward DL-structured RL model is constructed to predict the value function without the need for gradient backpropagation. To learn data-driven sparse representations with adaptive sparsity, we propose to use the learnable sparse regularizer ASR in the model, where the key hyperparameters of ASR can be trained to be adaptive to variable RL environments. To optimize the model efficiently, the model parameters are first pretrained in the pretraining stage, with only the value weights used for value function approximation needing to be fine-tuned for actual RL applications in the control training stage. Our comparative experiments in benchmark environments demonstrate that the proposed method can outperform existing state-of-the-art sparse-coding-based RL algorithms. In terms of accumulated rewards (used to measure th

关键词： Interference Reinforcement learning Adaptation models Training Heuristic algorithms Backpropagation Estimation Catastrophic interference deep reinforcement learning (DRL) dictionary learning (DL) learnable sparse regularizer sparse coding

来源：评论

学校读者我要写书评

暂无评论

A Contour Detection Method Based on the Projective coding Model of the Visual Cortex Information Flow

引用

IEEE TRANSACTIONS ON COGNITIVE AND DEVELOPMENTAL SYSTEMS 2024年第2期16卷 660-670页

作者： Cai, Zhefei Yang, Rui Fan, Yingle Wu, Wei Hangzhou Dianzi Univ Lab Pattern Recognit & Image Proc Hangzhou 310018 Peoples R China

Based on the information flow partition projection characteristics of P-type and M-type ganglion cells (referred to as P cells and M cells) and the neural sparse coding mechanism, this article proposed a new method for image contour detection. First, we considered the difference between M cells and P cells in detail sensitivity and the information transmission of different visual signals. The parallel visual pathway was constructed to simulate the prelevel characteristics of the V1 layer to obtain the primary contour response. Then, we introduced the orientation sensitivity and stimulus response difference of the visual receptive field to construct the visual information difference enhancement model. In consideration of the visual attention mechanism, we proposed an adaptive size sparse coding network model that simulates the prelevel characteristics of the V4 layer to intelligently focus the target contour features. At the same time, de-redundancy was performed to obtain the fine feature image. Finally, the hierarchical information feedback fusion was built, and the fine feature image was used to correct the primary contour response to obtain complete contour detection results. Taking the BSDS500 dataset as experimental objects, the results showed that the proposed method exhibits an effective tradeoff between contour extraction and texture suppression.

关键词： Contour detection information flow partition projection sparse coding visual attention mechanism visual receptive field

来源：评论

学校读者我要写书评

暂无评论

Efficient compression at the edge for real-time data acquisition in a billion-pixel X-ray camera

引用

NUCLEAR INSTRUMENTS & METHODS IN PHYSICS RESEARCH SECTION A-ACCELERATORS SPECTROMETERS DETECTORS AND ASSOCIATED EQUIPMENT 2024年 1058卷

作者： Rahali, Hamza Ezzaoui Rahimifar, Mohammad Mehdi Granger, Charles-Etienne Wang, Zhehui Therrien, Audrey C. Interdisciplinary Inst Technol Innovat 3IT 3000 Bd Univ Sherbrooke PQ J1K0A5 Canada Los Alamos Natl Lab Los Alamos NM 87545 USA

Recent advances in radiation detectors have significantly improved the resolution and frame rate of X-ray imaging systems at the cost of large data throughputs that can be challenging to transfer, store, and process. Such is the case of the billion-pixel X-ray camera, where the estimated throughput of 10 terabytes per second (TB/s) exceeds the capabilities of current data acquisition systems (DAQs) for real-time transfer and processing. To address this, we propose a lossy, multi-stage compression scheme to reduce data at the source. This work leverages machine learning (ML) to produce sparse representations of the camera's images, followed by quantization and entropy coding for compression. The proposed scheme achieves a 100:1 compression ratio with high PSNR and SSIM scores on X-ray source images. Finally, we implement the sparse coding neural network (NN) on an FPGA to showcase the feasibility, low power, and low latency of edge data compression.

关键词： X-ray instrumentation Signal processing Data compression sparse coding Neural networks Edge computing

来源：评论

学校读者我要写书评

暂无评论

A Pre-Voiding Alarm System Using Wearable Ultrasound and Machine Learning Algorithms for Children With Nocturnal Enuresis

引用

IEEE JOURNAL OF TRANSLATIONAL ENGINEERING IN HEALTH AND MEDICINE 2024年 12卷 643-658页

作者： Wang, Jun Dai, Zeyang Liu, Xiao Fudan Univ Sch Informat Sci & Technol Shanghai 200433 Peoples R China Fudan Univ State Key Lab Integrated Chips & Syst Shanghai 201203 Peoples R China Greater Bay Area Inst Precis Med Guangzhou 511400 Peoples R China

Nocturnal enuresis is a bothersome condition that affects many children and their caregivers. Post-voiding systems is of little value in training a child into a correct voiding routing while existing pre-voiding systems suffer from several practical limitations, such as cumbersome hardware, assuming individual bladder shapes being universal, and being sensitive to sensor placement error. Methods: A low-voltage ultrasound system with machine learning has been developed in estimating bladder filling status. A custom-made flexible 1D transducer array has been excited by low-voltage coded pulses with a pulse compression technique for an enhanced signal-to-noise ratio. In order to minimize the negative influence of possible transducer misplacement, a multiple-position training strategy using machine learning has been adopted in this work. Three popular classification methods, KNN, SVM and sparse coding, have been utilized to classify the acquired different volumes ranging from 100 ml to 300 ml into two categories: low volume and high volume. The low-volume category requires no further action while the high-volume category triggers an alarm to alert the child and caregiver. Results: When the sensor placement is ideal, i.e., the position of the practical sensor placement is on spot with the trained position, the precision and recall of the classification using sparse coding are 0.957 +/- 0.02 and 0.958 +/- 0.02 , respectively. Even if the transducer array is misplaced by up to 4.5 mm away from the ideal location, the proposed system is able to maintain high classification accuracy (precision >= 0.75 and recall >= 0.75 ). Category: Early/Pre-Clinical Research Clinical and Translational Impact: The proposed ultrasound sensor system for nocturnal enuresis is of significant clinical and translational value as it addresses two major issues that limit the wide adoption of similar devices. Firstly, it offers enhanced safety as the entire system has been ...

关键词： Bladder Transducers Shape Accuracy Ultrasonic imaging Pediatrics Monitoring Bladder filling status bladder volume machine learning sparse coding ultrasound system wearable system

来源：评论

学校读者我要写书评

暂无评论

Hessian sparse coding

引用

NEUROCOMPUTING 2014年 123卷 247-254页

作者： Zheng, Miao Bu, Jiajun Chen, Chun Zhejiang Univ Coll Comp Sci & Technol Hangzhou Zhejiang Peoples R China

sparse coding has received an increasing amount of interest in recent years. It finds a basis set that captures high-level semantics in the data and learns sparse coordinates in terms of the basis set. However, most of the existing approaches fail to consider the geometrical structure of the data space. Recently, a graph regularized sparse coding (GraphSC) is proposed to learn the sparse representations that explicitly take into account the local manifold structure, which used graph Laplacian as a smooth operator. However, the GraphSC based on graph Laplacian suffers from the fact that sparse coordinates are biased toward a constant and the Laplacian embedding often cannot preserve local topology well as we expected. In this paper, we propose a novel sparse coding algorithm called Hessian sparse coding (HessianSC). HessianSC is based on the second-order Hessian energy, which favors functions whose values vary linearly with respect to geodesic distance. HessianSC can overcome the drawbacks of Laplacian based methods. We show that our algorithm results in significantly improved performance when applied to image clustering task. (C) 2013 Elsevier B.V. All rights reserved.

关键词： sparse coding Manifold learning Hessian energy Image clustering

来源：评论

学校读者我要写书评

暂无评论

A Truncated EM Approach for Spike-and-Slab sparse coding

引用

JOURNAL OF MACHINE LEARNING RESEARCH 2014年第8期15卷 2653-2687页

作者： Sheikh, Abdul-Saboor Shelton, Jacquelyn A. Luecke, Joerg Tech Univ Berlin Fac Elect Engn & Comp Sci D-10587 Berlin Germany Carl von Ossietzky Univ Oldenburg D-26115 Oldenburg Germany

We study inference and learning based on a sparse coding model with 'spike-and-slab' prior. As in standard sparse coding, the model used assumes independent latent sources that linearly combine to generate data points. However, instead of using a standard sparse prior such as a Laplace distribution, we study the application of a more flexible 'spike-and-slab' distribution which models the absence or presence of a source's contribution independently of its strength if it contributes. We investigate two approaches to optimize the parameters of spike-and-slab sparse coding: a novel truncated EM approach and, for comparison, an approach based on standard factored variational distributions. The truncated approach can be regarded as a variational approach with truncated posteriors as variational distributions. In applications to source separation we find that both approaches improve the state-of-the-art in a number of standard benchmarks, which argues for the use of 'spike-and-slab' priors for the corresponding data domains. Furthermore, we find that the truncated EM approach improves on the standard factored approach in source separation tasks-which hints to biases introduced by assuming posterior independence in the factored variational approach. Likewise, on a standard benchmark for image denoising, we find that the truncated EM approach improves on the factored variational approach. While the performance of the factored approach saturates with increasing numbers of hidden dimensions, the performance of the truncated approach improves the state-of-the-art for higher noise levels.

关键词： sparse coding spike-and-slab distributions approximate EM variational Bayes unsupervised learning source separation denoising

来源：评论

学校读者我要写书评

暂无评论

m-BRe: New Variants for the Backward Replacement Approach to Enhance the Compression Capabilities of the Forward Greedy Pursuit

引用

IEEE ACCESS 2024年 12卷 138646-138662页

作者： Omara, A. N. Alotaibi, Abdullah Shawan Elect Res Inst Comp & Syst Dept Cairo 4473221 Egypt Shaqra Univ Comp Sci Dept Shaqra 11911 Saudi Arabia Ahram Canadian Univ Fac Comp Sci & Informat Technol Giza 3221405 Egypt

This paper introduces two variants of the backward replacement approach, T-BRe and H-BRe, aimed at enhancing the compression capabilities of forward greedy pursuit algorithms. The effectiveness of these methods is evaluated using three greedy pursuit algorithms (OMP, OOMP, and FBP) through simulations conducted on image and speech datasets with a mathematically generated dictionary. Additionally, a mathematical analysis of the H-BRe algorithm is provided to validate its effectiveness and address inherent limitations. Both analysis and simulations confirm that the H-BRe algorithm significantly improves the compression capabilities of forward greedy pursuit algorithms. The degree of enhancement varies depending on factors such as the forward modeling technique, stopping criterion, and the stochastic nature of the backward replacements. The high-performance (H) variants demonstrate superior performance across various modeling approaches, as evidenced by metrics such as PESQ and SSIM values, while the B and T variants also exhibit efficiency in specific scenarios. Despite these promising results, our methods encounter limitations such as time complexity, local error minimization, dependence on atom types, and the assumption of linear distribution. Future research directions include optimizing time complexity, balancing compression-quality trade-offs, globally minimizing replacement errors, exploring diverse atom types, and refining mathematical assumptions to further elucidate H-BRe's performance characteristics.

关键词： Matching pursuit algorithms Vectors Symmetric matrices Indexes Data compression sparse matrices Signal processing algorithms Greedy algorithms Backward elimination backward replacement data compression forward greedy pursuit 2D and cubic distortion matrix sparse coding

来源：评论

学校读者我要写书评

暂无评论

Double-constrained structured discriminant analysis-synthesis dictionary pair learning for pattern classification

引用

MULTIMEDIA TOOLS AND APPLICATIONS 2024年第10期83卷 30277-30295页

作者： Du, Haishun Zhang, Yonghao Wang, Yuxi He, Linbing Henan Univ Sch Artificial Intelligence Zhengzhou 450046 Peoples R China

Existing discriminant analysis-synthesis dictionary pair learning (ASDPL) methods learn a structured analysis dictionary containing multiple class-specific analysis sub-dictionaries and a structured synthesis dictionary containing multiple class-specific synthesis sub-dictionaries. Although existing discriminant ASDPL methods achieve promising results in the field of pattern classification, most of them ignore the correlation between an analysis sub-dictionary and a synthesis sub-dictionary that belong to different classes, which may degrade the discriminative ability of their learned dictionary pairs. Moreover, most existing discriminant ASDPL methods do not give an explicit constraint to ensure that the reconstruction error of training samples under the joint action of a structured analysis dictionary and a structured synthesis dictionary is as small as possible, leading to insufficient representational ability of their learned dictionary pairs. To address these issues, we present a double-constrained structured discriminant analysis-synthesis dictionary pair learning (DCSDDPL) method. Specifically, we first design a class-specific analysis-synthesis sub-dictionary pair reconstruction constraint term to ensure that the reconstruction error of training samples of a class is as small as possible under the joint action of the analysis and synthesis sub-dictionaries belonging to the same class, which helps to improve the representational ability of the learned dictionary pair. Then, we design an analysis-synthesis sub-dictionary pair independence constraint term to eliminate the correlation between an analysis sub-dictionary and a synthesis sub-dictionary belonging to different classes so as to ensure the discriminative ability of the learned dictionary pair. Finally, we formulate the DCSDDPL model by integrating the two constraint terms into the basic discriminant analysis-synthesis dictionary pair learning model. Moreover, we design an optimization algorithm and use

关键词： Pattern classification sparse coding Discriminant dictionary learning Discriminant dictionary pair learning

来源：评论

学校读者我要写书评

暂无评论

Pivotal Auto-Encoder via Self-Normalizing ReLU

引用

IEEE TRANSACTIONS ON SIGNAL PROCESSING 2024年 72卷 3201-3212页

作者： Goldenstein, Nelson Sulam, Jeremias Romano, Yaniv Technion Dept Elect & Comp Engn IL-3200003 Haifa Israel Johns Hopkins Univ Dept Biomed Engn Baltimore MD 21218 USA Johns Hopkins Univ Math Inst Data Sci Baltimore MD 21218 USA Technion Dept Comp Sci IL-3200003 Haifa Israel

sparse auto-encoders are useful for extracting low-dimensional representations from high-dimensional data. However, their performance degrades sharply when the input noise at test time differs from the noise employed during training. This limitation hinders the applicability of auto-encoders in real-world scenarios where the level of noise in the input is unpredictable. In this paper, we formalize single hidden layer sparse auto-encoders as a transform learning problem. Leveraging the transform modeling interpretation, we propose an optimization problem that leads to a predictive model invariant to the noise level at test time. In other words, the same pre-trained model is able to generalize to different noise levels. The proposed optimization algorithm, derived from the square root lasso, is translated into a new, computationally efficient auto-encoding architecture. After proving that our new method is invariant to the noise level, we evaluate our approach by training networks using the proposed architecture for denoising tasks. Our experimental results demonstrate that the trained models yield a significant improvement in stability against varying types of noise compared to commonly used architectures.

关键词： Transforms Noise level Noise Encoding Training sparse approximation Computational modeling sparse coding transform learning sparse auto-encoders square root lasso

来源：评论

学校读者我要写书评

暂无评论

Recognizing architecture styles by hierarchical sparse coding of blocklets

引用

INFORMATION SCIENCES 2014年 254卷 141-154页

作者： Zhang, Luming Song, Mingli Liu, Xiao Sun, Li Chen, Chun Bu, Jiajun Zhejiang Univ Coll Comp Sci Hangzhou 310027 Peoples R China

In this work, we propose a novel architecture style recognition model by introducing blocklets that capture the morphological characteristics of buildings. First, we decompose a building image into a collection of blocks, each representing a basic architecture component such as a stone pillar. To exploit the spatial correlations among blocks, we obtain locklets by extracting spatially adjacent blocks, and further formulate architecture style recognition as matching between blocklets extracted from different buildings. Toward an efficient blocklet-to-blocklet matching, a hierarchical sparse coding algorithm is proposed to represent each blocklet by a linear combination of basis blocklets. On the other hand, toward an effective matching process, an LDA [25,1]-like scheme is adopted to select the blocklets with high discrimination. Finally, we carry out architecture style recognition based on the selected highly discriminative blocklets. Experimental results on our own compiled data set demonstrate that the proposed approach outperforms several state-ofthe-art place/building recognition models. (C) 2013 Elsevier Inc. All rights reserved.

关键词： Architecture style Block let Hierarchical sparse coding

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：