Adversarial Erasing method based on graph neural network

Semantic segmentation is a traditional task that requires a large number of pixel-level ground truth label data sets, which is time-consuming and expensive. Recent developments in weakly-supervised settings have shown that reasonable performance can be obtained using only image-level labels. Classification is often used as an agent task to train deep neural networks and extract attention maps from them. The classification task only needs less supervision information to obtain the most discriminative part of the object. For this purpose, we propose a new end-to-end counter-wipe network. Compared with the baseline network, we propose a method to apply the graph neural network to obtain the first CAM. It is proposed to train the joint loss function to avoid the network weight sharing and cause the network to fall into a saddle point. Our experiments on the Pascal VOC2012 dataset show that 64.9% segmentation performance is obtained, which is an improvement of 2.1% compared to our baseline.

Hyperspace of finite unions of convergent sequences

The symbol S(X) denotes the hyperspace of finite unions of convergent sequences in a Hausdor˛ space X. This hyper-space is endowed with the Vietoris topology. First of all, we give a characterization of convergent sequence in S(X). Then we consider some cardinal invariants on S(X), and compare the character, the pseudocharacter, the sn-character, the so-character, the network weight and cs-network weight of S(X) with the corresponding cardinal function of X. Moreover, we consider rank k-diagonal on S(X), and give a space X with a rank 2-diagonal such that S(X) does not Gδ -diagonal. Further, we study the relations of some generalized metric properties of X and its hyperspace S(X). Finally, we pose some questions about the hyperspace S(X).

The Regional and Local Scale Evolution of the Spatial Structure of High-Speed Railway Networks—A Case Study Focused on Beijing-Tianjin-Hebei Urban Agglomeration

China has entered an era of rapid high-speed railway (HSR) development and the spatial structure of urban agglomerations will evolve in parallel with the development and evolution of the spatial structure of the HSR network. In this study, we explore how the spatial structure of an HSR network evolves at regional and local scales. Existing research into HSR network structures has mostly been carried out at a regional scale, and has therefore failed to reveal the spatial connections within a city. In this work, we progress the science by exploring it at a local scale. To describe the HSR network more accurately, we use the dwell time to simulate the passenger flow between stations and use the simulated passenger flow as the network weight. We use complex network analysis to investigate the evolution of the network’s spatial structure. Our results present the evolution of station locations, of community structure, and of the locations of connections between stations at a regional scale, and also show how HSR network development within core cities has impacted structures and connectivity at a local scale. These results help us to understand the spatial structure of urban agglomerations and cities, and provide evidence that can be used to optimize the structure of the HSR network within regions and cities.

Genetic Algorithmised Neuro Fuzzy System for Forecasting the Online Journal Visitors

Artificial Neural Network (ANN) is recognized as one of effective forecasting engines for various business fields. This approach fits well with non-linear data. In fact, it is a black box system with random weighting, which is hard to train. One way to improve its performance is by hybridizing ANN with other methods. In this paper, a hybrid approach, Genetic Algorithm-Neural Fuzzy System (GA-NFS) is proposed to predict the number of unique visitors of an online journal website. The neural network weight is precisely determined using GA. Afterwards, the best weight has been used for testing data and processed using Sugeno Fuzzy Inference System (FIS) for time-series forecasting. Based on experiment, GA-NFS have been produced accuracy with 0.989 of root mean square error (RMSE) that is lower than the RMSE of a common NFS (2,004). This may indicate that the GA based weighting is able to improve the NFS performance on forecasting the number of journal unique visitors.

Automated white corpuscles nucleus segmentation using deep neural network from microscopic blood smear

The white corpuscles nucleus segmentation from microscopic blood images is major steps to diagnose blood-related diseases. The perfect and speedy segmentation system assists the hematologists to identify the diseases and take appropriate decision for better treatment. Therefore, fully automated white corpuscles nucleus segmentation model using deep convolution neural network, is proposed in the present study. The proposed model uses the combination of ‘binary_cross_entropy’ and ‘adam’ for maintaining learning rate in each network weight. To validate the potential and capability of the above proposed solution, ALL-IDB2 dataset is used. The complete set of images is partitioned into training and testing set and tedious experimentations have been performed. The best performing model is selected and the obtained training and testing accuracy of best performing model is reported as 98.69 % and 99.02 %, respectively. The staging analysis of proposed model is evaluated using sensitivity, specificity, Jaccard index, dice coefficient, accuracy and structure similarity index. The capability of proposed model is compared with performance of the region-based contour and fuzzy-based level-set method for same set of images and concluded that proposed model method is more accurate and effective for clinical purpose.

Online learning for orientation estimation during translation in an insect ring attractor network

Insect neural systems are a promising source of inspiration for new algorithms for navigation, especially on low size, weight, and power platforms. There have been unprecedented recent neuroscience breakthroughs with Drosophila in behavioral and neural imaging experiments as well as the mapping of detailed connectivity of neural structures. General mechanisms for learning orientation in the central complex (CX) of Drosophila have been investigated previously; however, it is unclear how these underlying mechanisms extend to cases where there is translation through an environment (beyond only rotation), which is critical for navigation in robotic systems. Here, we develop a CX neural connectivity-constrained model that performs sensor fusion, as well as unsupervised learning of visual features for path integration; we demonstrate the viability of this circuit for use in robotic systems in simulated and physical environments. Furthermore, we propose a theoretical understanding of how distributed online unsupervised network weight modification can be leveraged for learning in a trajectory through an environment by minimizing of orientation estimation error. Overall, our results here may enable a new class of CX-derived low power robotic navigation algorithms and lead to testable predictions to inform future neuroscience experiments.

Ranking Cloud Computing Criteria in Developing Electronic Communications Services Using MCDM

According to the results and considering fuzzy calculations related to the capabilities of cloud computing in developing electronic communications services, the most important criteria in the "IT management in steel industries" cluster having (A) network code was "communicating with steel industries` costumers" having (AB) network code and fuzzy network weight equal to 0.096; the most important criteria in "cloud computing capabilities" cluster having (B) network code were "reducing steel industries` costs" having (BA) network code and fuzzy network weight equal to 0.191; and "providing rapid services to steel industries` costumers" having (BB) network code and fuzzy network weight equal to 0.120. on the other hand, the most important criteria in "developing electronic communications services" cluster having (C) network code was "storing the data in electronic communications services" having (CD) network code and fuzzy network weight equal to 0.123, since based on fuzzy logic calculation, they had the highest fuzzy rank in Matlab programming environment regarding cloud computing capabilities in developing electronic communications services.

Quantum-Inspired Evolutionary Algorithms for Neural Network Weight Distribution

Parkinson’s Disease is a degenerative neurological disorder with unknown origins, making it impossible to be cured or even diagnosed. The following article presents a Three-Layered Perceptron Neural Network model that is trained using a variety of evolutionary as well as quantum-inspired evolutionary algorithms for the classification of Parkinson's Disease. Optimization algorithms such as Particle Swarm Optimization, Artificial Bee Colony Algorithm and Bat Algorithm are studied along with their quantum-inspired counter-parts in order to identify the best suited algorithm for Neural Network Weight Distribution. The results show that the quantum-inspired evolutionary algorithms perform better under the given circumstances, with qABC offering the highest accuracy of about 92.3%. The presented model can be used not only for disease diagnosis but is also likely to find its applications in various other fields as well.

On the Effectiveness of Bayesian AutoML methods for Physics Emulators

The adoption of Machine Learning (ML) for building emulators for complex physical processes has seen an exponential rise in the recent years. While neural networks are good function approximators, optimizing the hyper-parameters of the network to reach a global minimum is not trivial, and often needs human knowl- edge and expertise. In this light, automatic ML or autoML methods have gained large interest as they automate the process of network hyper-parameter tuning. In addition, Neural Architecture Search (NAS) has shown promising outcomes for improving model performance. While autoML methods have grown in popularity for image, text and other applications, their effectiveness for high-dimensional, complex scientific datasets remains to be investigated. In this work, a data driven emulator for turbulence closure terms in the context of Large Eddy Simulation (LES) models is trained using Artificial Neural Networks and an autoML frame- work based on Bayesian Optimization, incorporating priors to jointly optimize the hyper-parameters as well as conduct a full neural network architecture search to converge to a global minima, is proposed. Additionally, we compare the effect of using different network weight initializations and optimizers such as ADAM, SGDM and RMSProp, to explore the best performing setting. Weight and function space similarities during the optimization trajectory are investigated, and critical differences in the learning process evolution are noted and compared to theory. We observe ADAM optimizer and Glorot initialization consistently performs better, while RMSProp outperforms SGDM as the latter appears to have been stuck at a local minima. Therefore, this autoML BayesOpt framework provides a means to choose the best hyper-parameter settings for a given dataset.