scholarly journals Few-Shot Charge Prediction with Data Augmentation and Feature Augmentation

2021 ◽  
Vol 11 (22) ◽  
pp. 10811
Author(s):  
Peipeng Wang ◽  
Xiuguo Zhang ◽  
Zhiying Cao
Keyword(s):  
High Frequency ◽  
Data Augmentation ◽  
Feature Fusion ◽  
Low Frequency ◽  
Good Effect ◽  
Model Optimization ◽  
Convolutional Network ◽  
Proposed Model ◽  
Virtual Samples ◽  
Feature Augmentation

The task of charge prediction is to predict the charge based on the fact description. Existing methods have a good effect on the prediction of high-frequency charges, but the prediction of low-frequency charges is still a challenge. Moreover, there exist some confusing charges that have relatively similar fact descriptions, which can be easily misjudged. Therefore, we propose a model with data augmentation and feature augmentation for few-shot charge prediction. Specifically, the model takes the text description as the input and uses the Mixup method to generate virtual samples for data augmentation. Then, the charge information heterogeneous graph is introduced, and a novel graph convolutional network is designed to extract distinguishability features for feature augmentation. A feature fusion network is used to effectively integrate the charge graph knowledge into the fact to learn semantic-enhanced fact representation. Finally, the semantic-enhanced fact representation is used to predict the charge. In addition, based on the distribution of each charge, a category prior loss function is designed to increase the contribution of low-frequency charges to the model optimization. The experimental results on real-work datasets prove the effectiveness and robustness of the proposed model.

scholarly journals Brain Medical Image Fusion Based on Dual-Branch CNNs in NSST Domain

2020 ◽  
Vol 2020 ◽  
pp. 1-15
Author(s):  
Zhaisheng Ding ◽  
Dongming Zhou ◽  
Rencan Nie ◽  
Ruichao Hou ◽  
Yanyu Liu
Keyword(s):  
Image Fusion ◽  
High Frequency ◽  
Medical Image ◽  
Low Frequency ◽  
Structural Features ◽  
Functional Information ◽  
Convolutional Network ◽  
Structure Information ◽  
Medical Image Fusion ◽  
Research Challenge

Computed tomography (CT) images show structural features, while magnetic resonance imaging (MRI) images represent brain tissue anatomy but do not contain any functional information. How to effectively combine the images of the two modes has become a research challenge. In this paper, a new framework for medical image fusion is proposed which combines convolutional neural networks (CNNs) and non-subsampled shearlet transform (NSST) to simultaneously cover the advantages of them both. This method effectively retains the functional information of the CT image and reduces the loss of brain structure information and spatial distortion of the MRI image. In our fusion framework, the initial weights integrate the pixel activity information from two source images that is generated by a dual-branch convolutional network and is decomposed by NSST. Firstly, the NSST is performed on the source images and the initial weights to obtain their low-frequency and high-frequency coefficients. Then, the first component of the low-frequency coefficients is fused by a novel fusion strategy, which simultaneously copes with two key issues in the fusion processing which are named energy conservation and detail extraction. The second component of the low-frequency coefficients is fused by the strategy that is designed according to the spatial frequency of the weight map. Moreover, the high-frequency coefficients are fused by the high-frequency components of the initial weight. Finally, the final image is reconstructed by the inverse NSST. The effectiveness of the proposed method is verified using pairs of multimodality images, and the sufficient experiments indicate that our method performs well especially for medical image fusion.

scholarly journals Model involving gene inactivation in the generation of autosomal recessive mutants in mammalian cells in culture.

1982 ◽  
Vol 2 (9) ◽  
pp. 1126-1133 ◽  
Author(s):  
A E Simon ◽  
M W Taylor ◽  
W E Bradley ◽  
L H Thompson
Keyword(s):  
Chinese Hamster Ovary ◽  
High Frequency ◽  
Mammalian Cells ◽  
Chinese Hamster Ovary Cells ◽  
Chinese Hamster ◽  
Low Frequency ◽  
Two Dimensional ◽  
Wild Type ◽  
Proposed Model ◽  
Class 1

We present evidence for a two-step model for expression of the recessive phenotype at the diploid adenine phosphoribosyl transferase (aprt) locus in Chinese hamster ovary cells. This model proposes a high-frequency event leading to allelic inactivation and a low-frequency event leading to a structural alteration of the APRT protein. Either event can occur first, resulting in two types of heterozygous cells. The proposed model is based on analysis of Chinese hamster ovary presumptive aprt heterozygotes and APRT- mutants, derived by two different laboratories. The major class of heterozygotes (class 1) had approximately 50% parental APRT activity, 50% immunologically precipitable APRT protein, and only wild-type enzyme as based on two-dimensional gel electrophoresis and thermal inactivation studies. We propose that one allele at the aprt locus has been inactivated in these heterozygotes. APRT- mutants derived from any single class 1 heterozygote arose at a low frequency and contained either no immunologically detectable APRT protein or an APRT enzyme which was, in most cases, demonstrably altered. The second class of heterozygotes, consisting of two independent isolates, gave rise to APRT- cells at a high frequency (10(-3) to 10(-5). These heterozygous cell lines had 50% of parental APRT activity and only wild-type spot, or wild-type and an electrophoretic variant spot, on two-dimensional gels. These aprt heterozygotes appear to have arisen by mutation at one allele. APRT- mutants derived from either heterozygote of this class had all lost the wild-type activity, consistent with the proposed model.

scholarly journals Convolutional neural network for automated mass segmentation in mammography

2020 ◽  
Vol 21 (S1) ◽  
Author(s):  
Dina Abdelhafiz ◽  
Jinbo Bi ◽  
Reda Ammar ◽  
Clifford Yang ◽  
Sheida Nabavi
Keyword(s):  
Neural Network ◽  
Convolutional Neural Network ◽  
Data Augmentation ◽  
Contextual Information ◽  
Automatic Segmentation ◽  
Ground Truth ◽  
Test Accuracy ◽  
Convolutional Network ◽  
Proposed Model ◽  
The Mean

Abstract Background Automatic segmentation and localization of lesions in mammogram (MG) images are challenging even with employing advanced methods such as deep learning (DL) methods. We developed a new model based on the architecture of the semantic segmentation U-Net model to precisely segment mass lesions in MG images. The proposed end-to-end convolutional neural network (CNN) based model extracts contextual information by combining low-level and high-level features. We trained the proposed model using huge publicly available databases, (CBIS-DDSM, BCDR-01, and INbreast), and a private database from the University of Connecticut Health Center (UCHC). Results We compared the performance of the proposed model with those of the state-of-the-art DL models including the fully convolutional network (FCN), SegNet, Dilated-Net, original U-Net, and Faster R-CNN models and the conventional region growing (RG) method. The proposed Vanilla U-Net model outperforms the Faster R-CNN model significantly in terms of the runtime and the Intersection over Union metric (IOU). Training with digitized film-based and fully digitized MG images, the proposed Vanilla U-Net model achieves a mean test accuracy of 92.6%. The proposed model achieves a mean Dice coefficient index (DI) of 0.951 and a mean IOU of 0.909 that show how close the output segments are to the corresponding lesions in the ground truth maps. Data augmentation has been very effective in our experiments resulting in an increase in the mean DI and the mean IOU from 0.922 to 0.951 and 0.856 to 0.909, respectively. Conclusions The proposed Vanilla U-Net based model can be used for precise segmentation of masses in MG images. This is because the segmentation process incorporates more multi-scale spatial context, and captures more local and global context to predict a precise pixel-wise segmentation map of an input full MG image. These detected maps can help radiologists in differentiating benign and malignant lesions depend on the lesion shapes. We show that using transfer learning, introducing augmentation, and modifying the architecture of the original model results in better performance in terms of the mean accuracy, the mean DI, and the mean IOU in detecting mass lesion compared to the other DL and the conventional models.

scholarly journals Self-Recurrent Learning and Gap Sample Feature Synthesis-Based Object Detection Method

2021 ◽  
Vol 2021 ◽  
pp. 1-15
Author(s):  
Lvjiyuan Jiang ◽  
Haifeng Wang ◽  
Kai Yan ◽  
Chengjiang Zhou ◽  
Songlin Li ◽  
...  
Keyword(s):  
Deep Learning ◽  
Object Detection ◽  
Production Efficiency ◽  
Detection Method ◽  
Feature Fusion ◽  
Information Filtering ◽  
Construction Work ◽  
Convolutional Network ◽  
Features Fusion ◽  
Feature Augmentation

Object detection-based deep learning by using the looking and thinking twice mechanism plays an important role in electrical construction work. Nevertheless, the use of this mechanism in object detection produces some problems, such as calculation pressure caused by multilayer convolution and redundant features that confuse the network. In this paper, we propose a self-recurrent learning and gap sample feature fusion-based object detection method to solve the aforementioned problems. The network consists of three modules: self-recurrent learning-based feature fusion (SLFF), residual enhancement architecture-based multichannel (REAML), and gap sample-based features fusion (GSFF). SLFF detects objects in the background through an iterative convolutional network. REAML, which serves as an information filtering module, is used to reduce the interference of redundant features in the background. GSFF adds feature augmentation to the network. Simultaneously, our model can effectively improve the operation and production efficiency of electric power companies’ personnel and guarantee the safety of lives and properties.

Prediction of the sound reduction index of clay hollow brick walls

2020 ◽  
Vol 27 (2) ◽  
pp. 155-168 ◽  
Author(s):  
Nicola Granzotto ◽  
Antonino Di Bella ◽  
Edoardo Alessio Piana
Keyword(s):  
High Frequency ◽  
Low Frequency ◽  
Sound Insulation ◽  
Frequency Range ◽  
Proposed Model ◽  
Different Types ◽  
Classic Approach ◽  
Hollow Brick ◽  
Sound Reduction ◽  
Reduction Index

Clay hollow brick walls are still popular in building industry, but the prediction of their sound insulation properties is not straightforward due to their inhomogeneous and anisotropic characteristics. In this article, a classic approach has been used to determine the sound transmission coefficient of brick walls, assuming an orthotropic behaviour and deriving the mechanical and dynamic characteristics from datasheet information. Different types of walls with horizontal and vertical mortar joints have been analysed. Experimental measurements of the sound reduction index carried out according to ISO 10140-2 standard have been performed, and the resulting values are compared with the predictions in the proposed model. It was found that the sound reduction index can be fairly predicted in the low-frequency range and it is correctly predicted in the mass law region, whereas in the high-frequency range the inner block structure is responsible for a loss of performance which is difficult to predict with the analytical methods.

scholarly journals Study on mechanism and filter efficacy of AGO/IAGO in the frequency domain

2020 ◽  
Vol ahead-of-print (ahead-of-print) ◽  
Author(s):  
Changhai Lin ◽  
Zhengyu Song ◽  
Sifeng Liu ◽  
Yingjie Yang ◽  
Jeffrey Forrest
Keyword(s):  
Frequency Domain ◽  
High Frequency ◽  
Statistical Data ◽  
Low Frequency ◽  
Digital Signal ◽  
Good Effect ◽  
Data Sequence ◽  
Content Type ◽  
Noise Interference ◽  
Long Period

PurposeThe purpose of this paper is to analyze the mechanism and filter efficacy of accumulation generation operator (AGO)/inverse accumulation generation operator (IAGO) in the frequency domain.Design/methodology/approachThe AGO/IAGO in time domain will be transferred to the frequency domain by the Fourier transform. Based on the consistency of the mathematical expressions of the AGO/IAGO in the gray system and the digital filter in digital signal processing, the equivalent filter model of the AGO/IAGO is established. The unique methods in digital signal processing systems “spectrum analysis” of AGO/IAGO are carried out in the frequency domain.FindingsThrough the theoretical study and practical example, benefit of spectrum analysis is explained, and the mechanism and filter efficacy of AGO/IAGO are quantitatively analyzed. The study indicated that the AGO is particularly suitable to act on the system's behavior time series in which the long period parts is the main factor. The acted sequence has good effect of noise immunity.Practical implicationsThe AGO/IAGO has a wonderful effect on the processing of some statistical data, e.g. most of the statistical data related to economic growth, crop production, climate and atmospheric changes are mainly affected by long period factors (i.e. low-frequency data), and most of the disturbances are short-period factors (high-frequency data). After processing by the 1-AGO, its high frequency content is suppressed, and its low frequency content is amplified. In terms of information theory, this two-way effect improves the signal-to-noise ratio greatly and reduces the proportion of noise/interference in the new sequence. Based on 1-AGO acting, the information mining and extrapolation prediction will have a good effect.Originality/valueThe authors find that 1-AGO has a wonderful effect on the processing of data sequence. When the 1-AGO acts on a data sequence X, its low-pass filtering effect will benefit the information fluctuations removing and high-frequency noise/interference reduction, so the data shows a clear exponential change trends. However, it is not suitable for excessive use because its equivalent filter has poles at the non-periodic content. But, because of pol effect at zero frequency, the 1-AGO will greatly amplify the low-frequency information parts and suppress the high-frequency parts in the information at the same time.

scholarly journals Convolutional neural network for automated mass segmentation in mammography

2020 ◽  
Author(s):  
Dina Abdelhafiz ◽  
Jinbo Bi ◽  
Reda Ammar ◽  
Clifford Yang ◽  
Sheida Nabavi
Keyword(s):  
Neural Network ◽  
Convolutional Neural Network ◽  
Data Augmentation ◽  
Contextual Information ◽  
Automatic Segmentation ◽  
Ground Truth ◽  
Test Accuracy ◽  
Convolutional Network ◽  
Proposed Model ◽  
The Mean

AbstractBackgroundAutomatic segmentation and localization of lesions in mammogram (MG) images are challenging even with employing advanced methods such as deep learning (DL) methods. We developed a new model based on the architecture of the semantic segmentation U-Net model to precisely segment mass lesions in MG images. The proposed end-to-end convolutional neural network (CNN) based model extracts contextual information by combining low-level and high-level features. We trained the proposed model using huge publicly available databases, (CBIS-DDSM, BCDR-01, and INbreast), and a private database from the University of Connecticut Health Center (UCHC).ResultsWe compared the performance of the proposed model with those of the state-of-the-art DL models including the fully convolutional network (FCN), SegNet, Dilated-Net, original U-Net, and Faster R-CNN models and the conventional region growing (RG) method. The proposed Vanilla U-Net model outperforms the Faster R-CNN model significantly in terms of the runtime and the Intersection over Union metric (IOU). Training with digitized film-based and fully digitized MG images, the proposed Vanilla U-Net model achieves a mean test accuracy of 92.6%. The proposed model achieves a mean Dice coefficient index (DI) of 0.951 and a mean IOU of 0.909 that show how close the output segments are to the corresponding lesions in the ground truth maps. Data augmentation has been very effective in our experiments resulting in an increase in the mean DI and the mean IOU from 0.922 to 0.951 and 0.856 to 0.909, respectively.ConclusionsThe proposed Vanilla U-Net based model can be used for precise segmentation of masses in MG images. This is because the segmentation process incorporates more multi-scale spatial context, and captures more local and global context to predict a precise pixel-wise segmentation map of an input full MG image. These detected maps can help radiologists in differentiating benign and malignant lesions depend on the lesion shapes. We show that using transfer learning, introducing augmentation, and modifying the architecture of the original model results in better performance in terms of the mean accuracy, the mean DI, and the mean IOU in detecting mass lesion compared to the other DL and the conventional models.

Model involving gene inactivation in the generation of autosomal recessive mutants in mammalian cells in culture

1982 ◽  
Vol 2 (9) ◽  
pp. 1126-1133
Author(s):  
A E Simon ◽  
M W Taylor ◽  
W E Bradley ◽  
L H Thompson
Keyword(s):  
Chinese Hamster Ovary ◽  
High Frequency ◽  
Mammalian Cells ◽  
Chinese Hamster Ovary Cells ◽  
Chinese Hamster ◽  
Low Frequency ◽  
Two Dimensional ◽  
Wild Type ◽  
Proposed Model ◽  
Class 1

We present evidence for a two-step model for expression of the recessive phenotype at the diploid adenine phosphoribosyl transferase (aprt) locus in Chinese hamster ovary cells. This model proposes a high-frequency event leading to allelic inactivation and a low-frequency event leading to a structural alteration of the APRT protein. Either event can occur first, resulting in two types of heterozygous cells. The proposed model is based on analysis of Chinese hamster ovary presumptive aprt heterozygotes and APRT- mutants, derived by two different laboratories. The major class of heterozygotes (class 1) had approximately 50% parental APRT activity, 50% immunologically precipitable APRT protein, and only wild-type enzyme as based on two-dimensional gel electrophoresis and thermal inactivation studies. We propose that one allele at the aprt locus has been inactivated in these heterozygotes. APRT- mutants derived from any single class 1 heterozygote arose at a low frequency and contained either no immunologically detectable APRT protein or an APRT enzyme which was, in most cases, demonstrably altered. The second class of heterozygotes, consisting of two independent isolates, gave rise to APRT- cells at a high frequency (10(-3) to 10(-5). These heterozygous cell lines had 50% of parental APRT activity and only wild-type spot, or wild-type and an electrophoretic variant spot, on two-dimensional gels. These aprt heterozygotes appear to have arisen by mutation at one allele. APRT- mutants derived from either heterozygote of this class had all lost the wild-type activity, consistent with the proposed model.

Simulations of high-resolution images of amorphous silicon films

1986 ◽  
Vol 44 ◽  
pp. 544-545
Author(s):  
G. Y. Fan ◽  
J. M. Cowley
Keyword(s):  
Electron Microscope ◽  
High Frequency ◽  
Fourier Transformation ◽  
Amorphous Materials ◽  
Low Frequency ◽  
Chromatic Aberration ◽  
Structure Information ◽  
Frequency Components ◽  
High Resolution Images ◽  
Spatial Frequency Spectrum

It is well known that the structure information on the specimen is not always faithfully transferred through the electron microscope. Firstly, the spatial frequency spectrum is modulated by the transfer function (TF) at the focal plane. Secondly, the spectrum suffers high frequency cut-off by the aperture (or effectively damping terms such as chromatic aberration). While these do not have essential effect on imaging crystal periodicity as long as the low order Bragg spots are inside the aperture, although the contrast may be reversed, they may change the appearance of images of amorphous materials completely. Because the spectrum of amorphous materials is continuous, modulation of it emphasizes some components while weakening others. Especially the cut-off of high frequency components, which contribute to amorphous image just as strongly as low frequency components can have a fundamental effect. This can be illustrated through computer simulation. Imaging of a whitenoise object with an electron microscope without TF limitation gives Fig. 1a, which is obtained by Fourier transformation of a constant amplitude combined with random phases generated by computer.

SEM image sharpness analysis

1996 ◽  
Vol 54 ◽  
pp. 142-143
Author(s):  
M. T. Postek ◽  
A. E. Vladar
Keyword(s):  
High Frequency ◽  
Low Frequency ◽  
Quantitative Information ◽  
Primary Electron ◽  
Image Sharpness ◽  
Critical Dimensions ◽  
Sem Image ◽  
Frequency Changes ◽  
Electron Microscopes ◽  
Scanning Electron

Fully automated or semi-automated scanning electron microscopes (SEM) are now commonly used in semiconductor production and other forms of manufacturing. The industry requires that an automated instrument must be routinely capable of 5 nm resolution (or better) at 1.0 kV accelerating voltage for the measurement of nominal 0.25-0.35 micrometer semiconductor critical dimensions. Testing and proving that the instrument is performing at this level on a day-by-day basis is an industry need and concern which has been the object of a study at NIST and the fundamentals and results are discussed in this paper.In scanning electron microscopy, two of the most important instrument parameters are the size and shape of the primary electron beam and any image taken in a scanning electron microscope is the result of the sample and electron probe interaction. The low frequency changes in the video signal, collected from the sample, contains information about the larger features and the high frequency changes carry information of finer details. The sharper the image, the larger the number of high frequency components making up that image. Fast Fourier Transform (FFT) analysis of an SEM image can be employed to provide qualitiative and ultimately quantitative information regarding the SEM image quality.

Sign in / Sign up

Export Citation Format

Share Document