检索结果-内蒙古大学图书馆

Feature Selection for recognition of Online Handwritten Bangla Characters

neural PROCESSING LETTERS 2019年第3期50卷 2281-2304页

作者： Sen, Shibaprasad Mitra, Mridul Bhattacharyya, Ankan Sarkar, Ram Schwenker, Friedhelm Roy, Kaushik Future Inst Engn & Management Kolkata 700150 India Jadavpur Univ Kolkata 700032 India Univ Ulm Inst Neural Informat Proc D-89069 Ulm Germany West Bengal State Univ Kolkata 700126 India

Feature selection through optimization techniques provides an interesting approach to minimize computational time with enhanced prediction capability, and has better cognizance of data in any pattern recognition application. This paper is an extended version of previously published work (Sen et al. in: 7th iapr TC3 workshop on artificial neural networks in pattern recognition, Ulm, Germany, pp 246-256, 2016), where a quad-tree based image segmentation approach has been discussed to estimate some topological and shape based features (192-attributed) for the recognition of online handwritten Bangla characters. The previous work achieved a recognition accuracy of 98.5% on a database consisting of 10,000 handwritten Bangla characters. In this paper, parameters, used in the previous version during feature estimation, are tuned to improve the performance of the overall system. Thereafter, krill-herd a bio-inspired, meta-heuristic algorithm has been applied to find the optimal feature vector by reducing the dimension of the original feature vector. The reduced feature vector has been fed to Sequential Minimal Optimization classifier for the recognition of the same online handwritten Bangla character database used in previous work. It has been observed that result obtained with this optimal feature set is almost equivalent as the result produced by the entire feature vector.

关键词： Feature selection Krill-herd optimization Online handwriting recognition Bangla script Quad-tree based image segmentation

来源：评论

学校读者我要写书评

暂无评论

Cluster-based Sample Selection for Document Image Binarization 2

Cluster-based Sample Selection for Document Image Binarizati...

引用

15th iapr International Conference on Document Analysis and recognition (ICDAR) / 2nd workshop of Machine Learning (WML)

作者： Krantz, Amandus Westphal, Florian Blekinge Inst Technol Dept Comp Sci Karlskrona Sweden

ISBN: (纸本)9781728150543

The current state-of-the-art, in terms of performance, for solving document image binarization is training artificial neural networks on pre-labelled ground truth data. As such, it faces the same issues as other, more conventional, classification problems;requiring a large amount of training data. However, unlike those conventional classification problems, document image binarization involves having to either manually craft or estimate the binarized ground truth data, which can be error-prone and time-consuming. This is where sample selection, the act of selecting training samples based on some method or metric, might help. By reducing the size of the training dataset in such a way that the binarization performance is not impacted, the required time spent creating the ground truth is also reduced. This paper proposes a cluster-based sample selection method that uses image similarity metrics and the relative neighbourhood graph to reduce the underlying redundancy of the dataset. The method, implemented with affinity propagation and the structural similarity index, reduces the training dataset on average by 49.57% while reducing the binarization performance only by 0.55%.

关键词： document image binarization sample selection neural networks computer vision

来源：评论

学校读者我要写书评

暂无评论

Inference with artificial neural networks on Analog Neuromorphic Hardware 2nd

Inference with Artificial Neural Networks on Analog Neuromor...

引用

2nd International workshop on IoT Streams for Data-Driven Predictive Maintenance, IoT Streams 2020, and 1st International workshop on IoT, Edge, and Mobile for Embedded Machine Learning, ITEM 2020, co-located with ECML/PKDD 2020

作者： Weis, Johannes Spilger, Philipp Billaudelle, Sebastian Stradmann, Yannik Emmel, Arne Müller, Eric Breitwieser, Oliver Grübl, Andreas Ilmberger, Joscha Karasenko, Vitali Kleider, Mitja Mauch, Christian Schreiber, Korbinian Schemmel, Johannes Kirchhoff-Institute for Physics Ruprecht-Karls-Universität Heidelberg Heidelberg Germany

ISBN: (纸本)9783030667696

The neuromorphic BrainScaleS-2 ASIC comprises mixed-signal neurons and synapse circuits as well as two versatile digital microprocessors. Primarily designed to emulate spiking neural networks, the system can also operate in a vector-matrix multiplication and accumulation mode for artificial neural networks. Analog multiplication is carried out in the synapse circuits, while the results are accumulated on the neurons’ membrane capacitors. Designed as an analog, in-memory computing device, it promises high energy efficiency. Fixed-pattern noise and trial-to-trial variations, however, require the implemented networks to cope with a certain level of perturbations. Further limitations are imposed by the digital resolution of the input values (5 bit), matrix weights (6 bit) and resulting neuron activations (8 bit). In this paper, we discuss BrainScaleS-2 as an analog inference accelerator and present calibration as well as optimization strategies, highlighting the advantages of training with hardware in the loop. Among other benchmarks, we classify the MNIST handwritten digits dataset using a two-dimensional convolution and two dense layers. We reach 98.0% test accuracy, closely matching the performance of the same network evaluated in software. © 2020, Springer Nature Switzerland AG.

关键词： Neurons

来源：评论

学校读者我要写书评

暂无评论

Historical Document Synthesis With Generative Adversarial networks 2

Historical Document Synthesis With Generative Adversarial Ne...

引用

15th iapr International Conference on Document Analysis and recognition (ICDAR) / 2nd workshop of Machine Learning (WML)

作者： Pondenkandath, Vinaychandran Alberti, Michele Diatta, Michael Ingold, Rolf Liwicki, Marcus Univ Fribourg Document Image & Voice Anal Grp DIVA Fribourg Switzerland Lulea Univ Technol Machine Learning Grp Lulea Sweden

ISBN: (纸本)9781728150543

This work tackles a particular image-to-image translation problem, where the goal is to transform an image from a source domain (modern printed electronic document) to a target domain (historical handwritten document). The main motivation of this task is to generate massive synthetic datasets of "historic" documents which can be used for the training of document analysis systems. By completing this task, it becomes possible to consider the generation of a tremendous amount of synthetic training data using only one single deep learning algorithm. Existing approaches for synthetic document generation rely on heuristics, or 2D and 3D geometric transformation-functions and are typically targeted at degrading the document. We tackle the problem of document synthesis and propose to train a particular form of Generative Adversarial neural networks, to learn a mapping function from an input image to an output image. With several experiments, we show that our algorithm generates an artificial historical document image that looks like a real historical document - for expert and non-expert eyes - by transferring the "historical style" to the classical electronic document.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Evaluation of group convolution in lightweight deep networks for object classification 1

引用

3rd workshop on Face and Facial Expression recognition from Real World Videos, FFER 2018, and 2nd International workshop on Deep Learning for pattern recognition, DLPR 2018, held at the 24th International Conference on pattern recognition, ICPR 2018

作者： Das, Arindam Boulay, Thomas Yogamani, Senthil Ou, Shiping Valeo Chennai India Valeo Paris France Valeo Galway Ireland Valeo Beijing China

ISBN: (数字)9783030121778

ISBN: (纸本)9783030121761

Deploying a neural network model on a low-power embedded platform is a challenging task. In this paper, we present our study on the efficacy of aggregated residual transformation (defined in ResNeXt that secured 2nd place in the ILSVRC 2016 classification task) for lightweight deep networks. The major contributions to this paper include (i) evaluation of group convolution, (ii) study on the impact of skip connection and various width for lightweight deep network. Our extensive experiments on different topologies show that employing aggregated convolution operation followed by point-wise convolution degrades the accuracy significantly. Furthermore as per our study, skip connections are not a suitable candidate for smaller networks and width is an important attribute to magnify the accuracy. Our embedded friendly networks are tested on ImageNet 2012 dataset where 3D convolution is a better alternative to aggregated convolution because of the 10% improvement in classification accuracy. © Springer Nature Switzerland AG 2019.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Convolutional neural network-based classification of histopathological images affected by data imbalance 1

引用

作者： Koziarski, Michal Kwolek, Bogdan Cyganek, Boguslaw Department of Electronics AGH University of Science and Technology Al. Mickiewicza 30 Kraków30-059 Poland

ISBN: (数字)9783030121778

ISBN: (纸本)9783030121761

In this paper we experimentally evaluated the impact of data imbalance on the convolutional neural networks performance in the histopathological image recognition task. We conducted our analysis on the Breast Cancer Histopathological Database. We considered four phenomena associated with data imbalance: how does it affect classification performance, what strategies of preventing imbalance are suitable for histopathological data, how presence of imbalance affects the value of new observations, and whether sampling training data from a balanced distribution during data acquisition is beneficial if test data will remain imbalanced. The most important findings of our experimental analysis are the following: while high imbalance significantly affects the performance, for some of the metrics small imbalance. Sampling training data from a balanced distribution had a decremental effect, and we achieved a better performance applying a dedicated strategy of dealing with imbalance. Finally, not all of the traditional strategies of dealing with imbalance translate well to the histopathological image recognition setting. © Springer Nature Switzerland AG 2019.

关键词： Image classification

来源：评论

学校读者我要写书评

暂无评论

Improving Multi-scale Face recognition Using VGGFace2 2nd

Improving Multi-scale Face Recognition Using VGGFace2

引用

2nd International workshop on Recent Advances in Digital Security: Biometrics and Forensics, BioFor 2019, 1st International workshop on pattern recognition for Cultural Heritage, PatReCH 2019, 1st International workshop eHealth in the Big Data and Deep Learning Era, e-BADLE 2019, International workshop on Deep Understanding Shopper Behaviors and Interactions in Intelligent Retail Environments, DEEPRETAIL 2019 and Industrial session held at the 20th International Conference on Image Analysis and Processing, ICIAP 2019

作者： Massoli, Fabio Valerio Amato, Giuseppe Falchi, Fabrizio Gennaro, Claudio Vairo, Claudio ISTI-CNR via G. Moruzzi 1 Pisa56124 Italy

ISBN: (纸本)9783030307530

Convolutional neural networks have reached extremely high performances on the Face recognition task. These models are commonly trained by using high-resolution images and for this reason, their discrimination ability is usually degraded when they are tested against low-resolution images. Thus, Low-Resolution Face recognition remains an open challenge for deep learning models. Such a scenario is of particular interest for surveillance systems in which it usually happens that a low-resolution probe has to be matched with higher resolution galleries. This task can be especially hard to accomplish since the probe can have resolutions as low as 8, 16 and 24 pixels per side while the typical input of state-of-the-art neural network is 224. In this paper, we described the training campaign we used to fine-tune a ResNet-50 architecture, with Squeeze-and-Excitation blocks, on the tasks of very low and mixed resolutions face recognition. For the training process we used the VGGFace2 dataset and then we tested the performance of the final model on the IJB-B dataset;in particular, we tested the neural network on the 1:1 verification task. In our experiments we considered two different scenarios: (1) probe and gallery with same resolution;(2) probe and gallery with mixed resolutions. Experimental results show that with our approach it is possible to improve upon state-of-the-art models performance on the low and mixed resolution face recognition tasks with a negligible loss at very high resolutions. © 2019, Springer Nature Switzerland AG.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

Skin Lesions Classification: A Radiomics Approach with Deep CNN 2nd

Skin Lesions Classification: A Radiomics Approach with Deep ...

引用

作者： Piantadosi, Gabriele Bovenzi, Giampaolo Argenziano, Giuseppe Moscarella, Elvira Parmeggiani, Domenico Docimo, Ludovico Sansone, Carlo Dip. di Ingegneria Elettrica e delle Tecnologie dell’Informazione University of Naples Federico II Naples Italy Dip. di Salute Mentale e Fisica e Medicina Preventiva University of Campania Luigi Vanvitelli Caserta Italy Dip. di Scienze Mediche e Chirurgiche Avanzate University of Campania Luigi Vanvitelli Caserta Italy

ISBN: (纸本)9783030307530

Supporting the early diagnosis of skin cancer is crucial for the sake of any kind of treatment or surgery. This work proposes to improve the outcome of automatic diagnoses approaches by using an ensemble of pre-trained deep convolutional neural networks and a suitable voting strategy. Moreover, a novel patching approach has been deployed. The proposal has been fairly evaluated with the literature proposals demonstrating good preliminary results. © 2019, Springer Nature Switzerland AG.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Slide Screening of Metastases in Lymph Nodes via Conditional, Fully Convolutional Segmentation 2nd

Slide Screening of Metastases in Lymph Nodes via Conditional...

引用

作者： Gerard, Gianluca Piastra, Marco Dipartimento di Ingegneria Industriale e dell’Informazione Università degli Studi di Pavia Via Ferrata 5 Pavia27100 Italy

ISBN: (纸本)9783030307530

We assess the viability of applying a conditional algorithm to the segmentation of Whole Slide Images (WSI) for human histopathology. Our objective is designing a deep network for automatic screening of large sets sentinel lymph-nodes of WSIs to detect those worth inspecting by a pathologist. Ideally, such system should modify and correct its behavior based on a limited set of examples, to foster interactivity and the incremental tuning to specific diagnostic pipelines and clinical practices and, not the least, to alleviate the task of collecting a suitable annotated dataset for training. In contrast, ‘classical’ supervised techniques require a vast dataset upfront and their behavior cannot be adapted unless through extensive retraining. The approach presented here is based on conditional and fully convolutional networks, which can segment a query image by conditioning on a support set of sparsely annotated images, fed at inference time. We describe the target scenario, the architecture used, and we present some preliminary results of segmentation experiments conducted on the publicly-available Camelyon16 dataset. © 2019, Springer Nature Switzerland AG.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

On Reproducibility of Deep Convolutional neural networks Approaches 2nd

On Reproducibility of Deep Convolutional Neural Networks App...

引用

2nd International workshop on Reproducible Research in pattern recognition, RRPR 2018

作者： Piantadosi, Gabriele Marrone, Stefano Sansone, Carlo DIETI - University of Naples Federico II Naples Italy

ISBN: (纸本)9783030239862

Nowadays, Machine Learning techniques are more and more pervasive in several application fields. In order to perform an evaluation as reliable as possible, it is necessary to consider the reproducibility of these models both at training and inference time. With the introduction of Deep Learning (DL), the assessment of reproducibility became a critical issue due to heuristic considerations made at training time that, although improving the optimization performances of such complex models, can result in non-deterministic outcomes and, therefore, not reproducible models. The aim of this paper is to quantitatively highlight the reproducibility problem of DL approaches, proposing to overcome it by using statistical considerations. We show that, even if the models generated by using several times the same data show differences in the inference phase, the obtained results are not statistically different. In particular, this short paper analyzes, as a case study, our ICPR2018 DL based approach for the breast segmentation in DCE-MRI, demonstrating the reproducibility of the reported results. © Springer Nature Switzerland AG 2019.

关键词： Deep neural networks

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：