Painter Classification Over the Novel Art Painting Data Set via The Latest Deep Neural Networks

This paper presents a systematic approach for accurate short-time cloud coverage prediction based on a machine learning (ML) approach. Based on a newly built omnidirectional ground-based sky camera system, local training and evaluation data sets were created. These were used to train several state-of-the-art deep neural networks for object detection and segmentation. For this purpose, the camera-generated a full hemispherical image every 30 min over two months in daylight conditions with a fish-eye lens. From this data set, a subset of images was selected for training and evaluation according to various criteria. Deep neural networks, based on the two-stage R-CNN architecture, were trained and compared with a U-net segmentation approach implemented by CloudSegNet. All chosen deep networks were then evaluated and compared according to the local situation.

Download Full-text

Modified Deep Neural Networks for Dog Breeds Identification

10.20944/preprints201812.0232.v1 ◽

2018 ◽

Cited By ~ 1

Author(s):

Aydin Ayanzadeh ◽

Sahand Vahidnia

Keyword(s):

Neural Networks ◽

Deep Neural Networks ◽

State Of The Art ◽

The State ◽

Fine Tuning ◽

Test Accuracy ◽

Data Sets ◽

Data Set

In this paper, we leverage state of the art models on Imagenet data-sets. We use the pre-trained model and learned weighs to extract the feature from the Dog breeds identification data-set. Afterwards, we applied fine-tuning and dataaugmentation to increase the performance of our test accuracy in classification of dog breeds datasets. The performance of the proposed approaches are compared with the state of the art models of Image-Net datasets such as ResNet-50, DenseNet-121, DenseNet-169 and GoogleNet. we achieved 89.66% , 85.37% 84.01% and 82.08% test accuracy respectively which shows thesuperior performance of proposed method to the previous works on Stanford dog breeds datasets.

Download Full-text

Mimicry Embedding Facilitates Advanced Neural Network Training for Image-Based Pathogen Detection

mSphere ◽

10.1128/msphere.00836-20 ◽

2020 ◽

Vol 5 (5) ◽

Author(s):

Artur Yakimovich ◽

Moona Huttunen ◽

Jerzy Samolej ◽

Barbara Clough ◽

Nagisa Yoshida ◽

...

Keyword(s):

Neural Network ◽

Neural Networks ◽

Deep Learning ◽

Deep Neural Networks ◽

Network Evolution ◽

Great Promise ◽

Data Sets ◽

Imaging Data ◽

Data Set ◽

Novel Strategy

ABSTRACT The use of deep neural networks (DNNs) for analysis of complex biomedical images shows great promise but is hampered by a lack of large verified data sets for rapid network evolution. Here, we present a novel strategy, termed “mimicry embedding,” for rapid application of neural network architecture-based analysis of pathogen imaging data sets. Embedding of a novel host-pathogen data set, such that it mimics a verified data set, enables efficient deep learning using high expressive capacity architectures and seamless architecture switching. We applied this strategy across various microbiological phenotypes, from superresolved viruses to in vitro and in vivo parasitic infections. We demonstrate that mimicry embedding enables efficient and accurate analysis of two- and three-dimensional microscopy data sets. The results suggest that transfer learning from pretrained network data may be a powerful general strategy for analysis of heterogeneous pathogen fluorescence imaging data sets. IMPORTANCE In biology, the use of deep neural networks (DNNs) for analysis of pathogen infection is hampered by a lack of large verified data sets needed for rapid network evolution. Artificial neural networks detect handwritten digits with high precision thanks to large data sets, such as MNIST, that allow nearly unlimited training. Here, we developed a novel strategy we call mimicry embedding, which allows artificial intelligence (AI)-based analysis of variable pathogen-host data sets. We show that deep learning can be used to detect and classify single pathogens based on small differences.

Download Full-text

Comparative Analysis of Deep Neural Networks for the Detection and Decoding of Data Matrix Landmarks in Cluttered Indoor Environments

Journal of Intelligent & Robotic Systems ◽

10.1007/s10846-021-01442-x ◽

2021 ◽

Vol 103 (1) ◽

Author(s):

Tiago Almeida ◽

Vitor Santos ◽

Oscar Martinez Mozos ◽

Bernardo Lourenço

Keyword(s):

Neural Networks ◽

Deep Learning ◽

Deep Neural Networks ◽

Input Image ◽

Automated Guided Vehicles ◽

Data Matrix ◽

Single Shot ◽

Indoor Environments ◽

Data Set ◽

Valid Solution

AbstractData Matrix patterns imprinted as passive visual landmarks have shown to be a valid solution for the self-localization of Automated Guided Vehicles (AGVs) in shop floors. However, existing Data Matrix decoding applications take a long time to detect and segment the markers in the input image. Therefore, this paper proposes a pipeline where the detector is based on a real-time Deep Learning network and the decoder is a conventional method, i.e. the implementation in libdmtx. To do so, several types of Deep Neural Networks (DNNs) for object detection were studied, trained, compared, and assessed. The architectures range from region proposals (Faster R-CNN) to single-shot methods (SSD and YOLO). This study focused on performance and processing time to select the best Deep Learning (DL) model to carry out the detection of the visual markers. Additionally, a specific data set was created to evaluate those networks. This test set includes demanding situations, such as high illumination gradients in the same scene and Data Matrix markers positioned in skewed planes. The proposed approach outperformed the best known and most used Data Matrix decoder available in libraries like libdmtx.

Download Full-text

Obtaining depth map from 2D non stereo images using deep neural networks

International Journal of Intelligent Unmanned Systems ◽

10.1108/ijius-03-2018-0007 ◽

2018 ◽

Vol 6 (3) ◽

pp. 134-146

Author(s):

Daniil Igorevich Mikhalchenko ◽

Arseniy Ivin ◽

Dmitrii Malov

Keyword(s):

Neural Networks ◽

Deep Neural Networks ◽

Depth Map ◽

Depth Information ◽

Data Sets ◽

Data Set ◽

Content Type ◽

Laser Sensors ◽

Depth Images ◽

Depth Prediction

Purpose Single image depth prediction allows to extract depth information from a usual 2D image without usage of special sensors such as laser sensors, stereo cameras, etc. The purpose of this paper is to solve the problem of obtaining depth information from 2D image by applying deep neural networks (DNNs). Design/methodology/approach Several experiments and topologies are presented: DNN that uses three inputs—sequence of 2D images from videostream and DNN that uses only one input. However, there is no data set, that contains videostream and corresponding depth maps for every frame. So technique of creating data sets using the Blender software is presented in this work. Findings Despite the problem of an insufficient amount of available data sets, the problem of overfitting was encountered. Although created models work on the data sets, they are still overfitted and cannot predict correct depth map for the random images, that were included into the data sets. Originality/value Existing techniques of depth images creation are tested, using DNN.

Download Full-text

A SELECTIVE LEARNING METHOD TO IMPROVE THE GENERALIZATION OF MULTILAYER FEEDFORWARD NEURAL NETWORKS

International Journal of Neural Systems ◽

10.1142/s0129065701000588 ◽

2001 ◽

Vol 11 (02) ◽

pp. 167-177 ◽

Cited By ~ 8

Author(s):

I. M. GALVÁN ◽

P. ISASI ◽

R. ALER ◽

J. M. VALLS

Keyword(s):

Neural Networks ◽

Learning Strategy ◽

Time Series Prediction ◽

Feedforward Neural Networks ◽

Training Data ◽

The Novel ◽

Learning Method ◽

Data Set ◽

Dynamic Selection

Multilayer feedforward neural networks with backpropagation algorithm have been used successfully in many applications. However, the level of generalization is heavily dependent on the quality of the training data. That is, some of the training patterns can be redundant or irrelevant. It has been shown that with careful dynamic selection of training patterns, better generalization performance may be obtained. Nevertheless, generalization is carried out independently of the novel patterns to be approximated. In this paper, we present a learning method that automatically selects the training patterns more appropriate to the new sample to be predicted. This training method follows a lazy learning strategy, in the sense that it builds approximations centered around the novel sample. The proposed method has been applied to three different domains: two artificial approximation problems and a real time series prediction problem. Results have been compared to standard backpropagation using the complete training data set and the new method shows better generalization abilities.

Download Full-text

Selectively Connected Self-Attentions for Semantic Role Labeling

Applied Sciences ◽

10.3390/app9081716 ◽

2019 ◽

Vol 9 (8) ◽

pp. 1716

Author(s):

Jaehui Park

Keyword(s):

Neural Networks ◽

Deep Neural Networks ◽

State Of The Art ◽

Hierarchical Structures ◽

Large Data ◽

Neural Model ◽

Semantic Role ◽

Semantic Role Labeling ◽

Data Set ◽

Training Time

Semantic role labeling is an effective approach to understand underlying meanings associated with word relationships in natural language sentences. Recent studies using deep neural networks, specifically, recurrent neural networks, have significantly improved traditional shallow models. However, due to the limitation of recurrent updates, they require long training time over a large data set. Moreover, they could not capture the hierarchical structures of languages. We propose a novel deep neural model, providing selective connections among attentive representations, which remove the recurrent updates, for semantic role labeling. Experimental results show that our model performs better in accuracy compared to the state-of-the-art studies. Our model achieves 86.6 F1 scores and 83.6 F1 scores on the CoNLL 2005 and CoNLL 2012 shared tasks, respectively. The accuracy gains are improved by capturing the hierarchical information using the connection module. Moreover, we show that our model can be parallelized to avoid the repetitive updates of the model. As a result, our model reduces the training time by 62 percentages from the baseline.

Download Full-text

Risk prediction with office and ambulatory blood pressure using artificial intelligence

10.1101/2020.01.17.20017798 ◽

2020 ◽

Author(s):

Pedro Guimarães ◽

Andreas Keller ◽

Michael Böhm ◽

Lucas Lauder ◽

José L. Ayala ◽

...

Keyword(s):

Artificial Intelligence ◽

Machine Learning ◽

Blood Pressure ◽

Neural Networks ◽

Ambulatory Blood Pressure ◽

Deep Neural Networks ◽

White Coat ◽

Risk Scores ◽

Support Vector ◽

The Novel

AbstractBackgroundTo develop and validate a novel, machine learning-derived model for prediction of cardiovascular (CV) mortality risk using office (OBP) and ambulatory blood pressure (ABP), to compare its performance with existing risk scores, and to assess the possibility of predicting ABP phenotypes (i.e. white-coat, ambulatory and masked hypertension) utilizing clinical variables.MethodsUsing data from 63,910 patients enrolled in the Spanish ABP monitoring registry, machine-learning approaches (logistic regression, support vector machine, gradient boosted decision trees, and deep neural networks) and stepwise forward feature selection were used for the classification of the data.ResultsOver a median follow-up of 4.7 years, 3,808 deaths occurred from which 1,295 were from CV causes. The performance for all tested classifiers increased while adding up to 10 features and converged thereafter. For the prediction of CV mortality, deep neural networks yielded the highest clinical performance. The novel mortality prediction models using OBP (CV-MortalityOBP) and ABP (CV-MortalityABP) outperformed all other risk scores. The area under the curve (AUC) achieved by the novel approach, using OBP variables only, was already significantly higher when compared with the AUC of Framingham score (0.685 vs 0.659, p = 1.97×10−22), the SCORE (0.679 vs 0.613, p = 6.21×10−22), and ASCVD (0.722 vs 0.639, p = 8.03×10−30) risk score. However, prediction of CV mortality with ABP instead of OBP data led to a significant increase in AUC (0.781 vs 0.752, p = 1.73×10−42), accuracy, balanced accuracy and sensitivity. The sensitivity and specificity for detection of ambulatory, masked, and white-coat hypertension ranged between 0.653-0.661 and 0.573-0.651, respectively.ConclusionWe developed a novel risk calculator for CV death using artificial intelligence based on a large cohort of patients included in the Spanish ABP monitoring registry. The receiver operating characteristic curves for CV-MortalityOBP and CV-MortalityABP with deep neural networks models outperformed all other risk metrics. Prediction of CV mortality using ABP data led to a significant increase in performance metrics. The prediction of ambulatory phenotypes using clinical characteristics, including OBP, was limited.

Download Full-text

DEPENDENCE OF THE INITIAL DISTRIBUTION OF WEIGHTS ON THE TRAINING EFFICIENCY OF DEEP NEURAL NETWORKS

Applied Mathematics and Fundamental Informatics ◽

10.25206/2311-4908-2020-7-4-24-33 ◽

2020 ◽

Vol 7 (4) ◽

pp. 024-033

Author(s):

I. P. Kolganov ◽

Keyword(s):

Neural Networks ◽

Regression Model ◽

Deep Neural Networks ◽

Initial Distribution ◽

Training Data ◽

Fast Method ◽

Data Set ◽

Lennard Jones ◽

Complex Models ◽

Training Efficiency

The article is devoted to the influence of the initial weights, during training. Data set is generated by computation methods based on the Lennard-Jones equation. The trained regression model predicts interaction energy between two particles. The sum square error is chosen as the criterion for evaluating of the efficiency. The article presents graphs of the efficiency indicator from the number of training steps, for different regimes. The data obtained in this article will allow to analyze the possibility of fast method designing for complex models.

Download Full-text

Book spine recognition with the use of deep neural networks

Computer Optics ◽

10.18287/2412-6179-co-731 ◽

2020 ◽

Vol 44 (6) ◽

pp. 968-977

Author(s):

M.O. Kalinina ◽

P.L. Nikolaev

Keyword(s):

Neural Network ◽

Neural Networks ◽

Network Performance ◽

Deep Neural Network ◽

Deep Neural Networks ◽

Original Data ◽

Network Architectures ◽

Data Set ◽

The Neural Network ◽

Processing Information

Nowadays deep neural networks play a significant part in various fields of human activity. Especially they benefit spheres dealing with large amounts of data and lengthy operations on obtaining and processing information from the visual environment. This article deals with the development of a convolutional neural network based on the YOLO architecture, intended for real-time book recognition. The creation of an original data set and the training of the deep neural network are described. The structure of the neural network obtained is presented and the most frequently used metrics for estimating the quality of the network performance are considered. A brief review of the existing types of neural network architectures is also made. YOLO architecture possesses a number of advantages that allow it to successfully compete with other models and make it the most suitable variant for creating an object detection network since it enables some of the common disadvantages of such networks to be significantly mitigated (such as recognition of similarly looking, same-color book coves or slanted books). The results obtained in the course of training the deep neural network allow us to use it as a basis for the development of the software for book spine recognition.

Download Full-text