Search

Although computer-aided diagnosis (CAD) have shown excellent performance in Breast cancer (BC) histopathological image, it commonly requires a high-level network, and the recognition efficiency is frequently unsatisfied due to the complex structure of histopathological image. In this study, a ten-layers convolutional neural network (CNN) model called "ColorDeep" is used to extract color features corresponding to the different tissue parts in cell, pure color image slices obtained by a three-channel separation and reconstruction method are used as model input. Two models are tested on the BreaKHis dataset show that images under four magnifications achieved the recognition accuracy of 96.89%-99.67% at the image level, which is better than many state-of-the-art methods. The characteristics contained by the B channel have the largest effect on BC recognition, and compared to other research results, the proposed model improves the recognition speed on a single image by about 0.1s. More importantly, instead of using large histopathological images to input into the model for BC diagnosis, and instead of segmenting the nuclei, only the reconstructed B-channel features containing the nuclei region in the stained BC image need to be input into the model to enable accurate diagnosis of BC.

We report the identification of a positive cross-correlation signal between the unresolved $\gamma$-ray emission, measured by the \emph{Fermi} Large Area Telescope, and four different galaxy cluster catalogues. The selected catalogues peak at low-redshift and span different frequency bands, including infrared, optical and X-rays. The signal-to-noise ratio of the detected cross-correlation amounts to 3.5 in the most significant case. We investigate and comment about its possible origin, in terms of compact $\gamma$-ray emission from AGNs inside clusters or diffuse emission from the intracluster medium. The analysis has been performed by introducing an accurate estimation of the cross-correlation power-spectrum covariance matrix, built with mock realisations of the gamma and galaxy cluster maps. Different methods to produce the mock realizations starting from the data maps have been investigated and compared, identifying suitable techniques which can be generalized to other cross-correlation studies., Comment: 24 pages, 17 figures

Maximum likelihood (ML) method for direction of arrival (DOA) estimation achieves an excellent performance in array signal processing, but the complexity and computational load of searching the multidimensional nonlinear function prevented it from practical application. Based on squirrel search algorithm (SSA), an improved SSA (ISSA) for ML DOA estimation is proposed in this paper, which can reduces the computational complexity. The idea of spatial variation and diffuse inspired by the invasive weed optimization(IWO) algorithm is applied to ISSA. The simulation experiments compared ISSA with SSA, IWO, seeker optimization algorithm(SOA), sine cosine algorithm (SCA), genetic algorithm (GA), particle swarm optimization (PSO) and differential evolution (DE) method for ML DOA estimator show that the proposed algorithm has faster convergence speed, fewer iterations and lower root mean square error(RMSE) under different number of signal sources, different signal to noise ratio(SNR) and different population size. Therefor the proposed algorithm does not only ensure the estimation accuracy, but also greatly reduce the computation complexity of multidimensional nonlinear optimization for the ML method. Finally, the test experiment using Micro Electronic Mechanical Systems(MEMS) vector hydrophone array in Fenhe lake show the engineering practicability of proposed ML DOA estimator with ISSA.The results obtained will be valuable in the application of engineering.

The fireworks algorithm, which is inspired by the explosion of fireworks, is a new swarm-based meta-heuristic algorithm for global optimization. This work proposes an improved fireworks optimization algorithm (IFWA) based on the enhanced fireworks algorithm (EFWA). Three aspects of improvement are presented after an analysis of the drawbacks of EFWA. These improvements are a new explosion scheme, GS-Gaussian explosion operator, and deep information exchange strategy. The proposed IFWA is tested on 23 benchmark function optimization problems and a real engineering problem, namely, optimal controller design for automotive active suspension. Optimization results prove that IFWA has competitive advantage compared with EFWA and other popular meta-heuristic algorithms and demonstrates the potential to solve real problems effectively.

This paper proposes a hybrid algorithm called ISSA based on the combination of squirrel search algorithm (SSA) proposed in 2019 and invasive weed optimization (IWO) proposed in 2006. About 36 benchmark functions are employed to test the performances of ISSA. Then, ISSA is combined with support vector machine (SVM) and deterministic maximum-likelihood (DML) algorithm, respectively, and the two corresponding models ISSA-SVM and ISSA-DML are established for performing the grade classifications of air quality and the direction of arrival (DOA) estimation of MEMS vector hydrophone, respectively. The results of 36 benchmark functions prove that the proposed ISSA is able to provide very competitive results in terms of the average values, the standard derivation, and the convergence curves. The average accuracy rate of classification of ISSA-SVM model is the best and reaches 87.91971%, and the DOA estimations of ISSA-DML have the least root mean square error (RMSE) and the closest to the actual angles. Therefore, it is concluded that the proposed ISSA is an effective algorithm for function optimizations and is suitable to be combined with other algorithms and machine learning for classification and estimation.

In this work we study the cross-correlation between Fermi-LAT diffuse γ-ray maps and galaxy cluster catalogues in order to search for a diffuse dark matter signal. We employ an accurate estimation of the error covariance matrix and we select 4 galaxy cluster catalogues at low-redshift 0 < z < 0.2 with large-halo mass M500 > 1013M⊙ including infrared, optical and X-rays wavebands. We consider weakly interacting massive particle (WIMP) dark matter with 100% branching ratios into τ+τ−, $b\bar{b}$, W+W− and μ+μ−. No evidence for a WIMP dark matter signal is identified, and we derive competitive bounds. The WIMP thermally averaged cross section is excluded at 95% C.L. for dark matter masses below 20 GeV and annihilation in the τ+τ− channel.

In this paper, we estimate the cross-correlation power spectra between the Planck 2018 cosmic microwave background (CMB) temperature anisotropy map and the unresolved γ -ray background (UGRB) from the 9-years Fermi-Large Area Telescope (LAT) data. In this analysis, we use up to nine energy bins over a wide energy range of [0.631, 1000] GeV from the Fermi-LAT UGRB data. Firstly, we find that the Fermi data with the energy ranges [1.202, 2.290] GeV and [17.38, 36.31] GeV show the positive evidence for the Integrated Sachs-Wolfe (ISW) effect at 1 . 8 σ confidence level, and the significance would be increased to 2 . 7 σ when using these two energy bins together. Secondly, we apply the single power-law model to normalize the amplitude and use all the nine Fermi energy bins to measure the significance of the ISW effect, we obtained A amp = 0 . 95 ± 0 . 53 (68% C.L.). For the robustness test, we implement a null hypothesis by randomizing the Fermi mock maps of nine energy bins and obtain the non-detection of ISW effect, which confirms that the ISW signal comes from the Fermi-LAT diffuse γ -ray data and is consistent with the standard Λ CDM model prediction essentially. We use a cross-correlation coefficient to show the relation between different energy bins. Furthermore, we vary the cut ranges | b | of galactic plane on the mask of Fermi map and carefully check the consequent influence on the ISW signal detection.

The architecture and parameter initialization of wavelet neural network are discussed and a novel initialization method is proposed. The new approach can be regarded as a dynamic clustering procedure which will derive the neuron number as well as the initial value of translation and dilation parameters according to the input patterns and the activating wavelets functions. Three simulation examples are given to examine the performance of our method as well as Zhang's heuristic initialization approach. The results show that the new approach not only can decide the WNN structure automatically, but also provides superior initial parameter values that make the optimization process more stable and quickly.

Based on gene expression, we have classified 53 colon cancer patients with UICC II into two groups: relapse and no relapse. Samples were taken from each patient, and gene information was extracted. Of the 53 samples examined, 500 genes were considered proper through analyses by S-Kohonen, BP, and SVM neural networks. Classification accuracy obtained by S-Kohonen neural network reaches 91%, which was more accurate than classification by BP and SVM neural networks. The results show that S-Kohonen neural network is more plausible for classification and has a certain feasibility and validity as compared with BP and SVM neural networks.

In order to ascertain the relationship between gene expression and colon cancer localization, a classification method based on random gene selection and a self-organizing map network is proposed. Different numbers of genes were selected randomly from 54,675 genes of 53 colon cancer patients in stage union for international cancer control II. These patients were then divided into two sets: a training set of 36 and a validation set of 17 patients. In this study, we randomly selected 1000, 100, 50, 30, 10, 5, and 3 genes, 1000 times, respectively. The minimum misclassification ratio of each gene group was 3/17 to 4/17, and the percentage of gene groups that were less than 0.25 was approximately 1-7%. Moreover, the misclassification ratio of most gene groups (about 82-89%) was lower than 0.4. Through the analysis of these low misclassification ratio gene groups, we found that there were few common genes between them. This revealed that colon cancer localization is not associated with a single gene group but with many gene groups. Furthermore, K-fold cross validation was used to test the reliability of the possible informative genes, and the results indicated that using gene expression to classify colon tumor localization was not feasible.

In this paper, a novel self-creating disk-cell-splitting (SCDCS) algorithm is proposed for training the radial wavelet neural network (RWNN) model. Combining with the least square (LS) method which determines the linear weight coefficients, SCDCS can create neurons adaptively on a disk according to the distribution of input data and learning goals. As a result, a disk map is made for input data as well as a RWNN model with proper architecture and parameters can be decided for the recognition task. The proposed SCDCS-LS based RWNN model is employed for the recognition of license plate characters. Compared to the classical radial-basis-function (RBF) network with K-means clustering and LS, the proposed model can make a better recognition performance even with fewer neurons.

An effective two-level self-organizing map (SOM) neural network for direction of arrival (DOA) of sound signals estimation is proposed. The approach is based on the distance difference of arrival (DDOA) and a uniform linear sensor array in a 2D plane; it performs a nonlinear mapping between the DDOA vectors and angles of arrival (AOA). We found that the topological order of DDOA vectors and AOAs of same signals is uniform; thus, the topological order preserving of SOM network makes it valid to estimate AOA through DDOA. From the results of simulations and lake experiments, it is shown that the network has the advantage of accuracy and robustness, can be trained in advance, and is easy to implement.

In this work, both experimental and theoretical study on the FT-IR and Raman spectra as well as 1 H NMR and 13 C NMR chemical shifts of 4,4′-dibromodiphenyl ether have been carried out. The optimized geometry was obtained by using both HF and density functional B3LYP method with the 6-31G(d) and 6-311+G(d, p) basis sets. The calculated bond lengths and dihedral angles for both methods on 6-31G(d) level show the best agreement with the experimental data, while the dihedral angles of C 1′ O C 1 C 6 and C 1′ O C 1 C 2 , critical geometry parameters for conformers in the ground state, indicates significant deviation of HF results from the experimental information. The harmonic vibration frequencies and intensities in IR and Raman spectra and chemical shifts of the molecule were calculated on the B3LYP/6-31G(d) and B3LYP/6-311+G(d, p) levels. The scaled theoretical vibration frequencies present good agreement with the experimental values. The larger basis set makes no significant improvement in the accuracy of the vibration frequencies. Besides, chemical shifts of hydrogen and carbon computed on B3LYP/6-31G(d) level agree well with the observations.