The detection of program vulnerabilities remains a challenging task in software security. The existing vulnerability detection methods rarely consider the multidimensional feature space complementarity of program graph structures, which easily overlooks contextual environment features and syntax structure features. This disadvantage leads to insufficient performance in capturing complex structural features, which hinders the improvement in detection accuracy. To address this issue, this paper introduces a novel vulnerability detection method, EnGS2F, which adopts the representation learning of an enhanced graph structure to improve the efficiency of capturing vulnerability information. On the dimension of the graph structure, a context relationship graph (CRG) is integrated on the basis of a program dependency graph (PDG) to enrich the global structural context representation. On the dimension of graph nodes, abstract syntax tree (AST) embedding and paragraph embedding are integrated to solve the problem of insufficient feature space complementarity. Moreover, the combination of a gated graph neural network (GGNN) with a graph attention mechanism further improves the learning performance of the enhanced graph structure. EnGS2F has been rigorously evaluated on program slices from open-source vulnerability datasets, demonstrating significant improvements over current competitive methods in detecting program vulnerabilities. Specifically, EnGS2F achieved a significant increase in the F1 score, outperforming existing technologies by 6%.

Background:Multi-source cross-project defect prediction (MSCPDP) attempts to transfer defect knowledge learned from multiple source projects to the target project. MSCPDP has drawn increasing attention from academic and industry communities owing to its advantages compared with single-source cross-project defect prediction (SSCPDP). However, two main problems, which are how to effectively extract the transferable knowledge from each source dataset and how to measure the amount of knowledge transferred from each source dataset to the target dataset, seriously restrict the performance of existing MSCPDP models.Objective:In this paper, we propose a novel multi-source transfer weighted ensemble learning (MASTER) method for MSCPDP.Method:MASTER measures the weight of each source dataset based on feature importance and distribution difference and then extracts the transferable knowledge based on the proposed feature-weighted transfer learning algorithm. Experiments are performed on 30 software projects. We compare MASTER with the latest state-of-the-art MSCPDP methods with statistical test in terms of famous effort-unaware measures (i.e., PD, PF, AUC, and MCC) and two widely used effort-aware measures (Popt20% and IFA).Result:The experiment results show that: 1) MASTER can substantially improve the prediction performance compared with the baselines, e.g., an improvement of at least 49.1% in MCC, 48.1% in IFA; 2) MASTER significantly outperforms each baseline on most datasets in terms of AUC, MCC,Popt20% and IFA; 3) MSCPDP model significantly performs better than the mean case of SSCPDP model on most datasets and even outperforms the best case of SSCPDP on some datasets.Conclusion:It can be concluded that 1) it is very necessary to conduct MSCPDP, and 2) the proposed MASTER is a more promising alternative for MSCPDP.

The analysis of high resolution (6m) radar data was carried out on the Antarctic sea ice segmentation. This paper presented two methods for Synthetic Aperture Radar (SAR) sea ice image segmentation: k-means clustering method and threshold value method (binary). First, the speckle noise in SAR image was filtered and segmented by two methods. Then the results were compared. It was found that both methods could distinguish the crack and small ice, and these two sea ice distribution results were almost similar. The difference between two methods was: image of threshold value method contained disordered spots and isolated small ice, while k-mean method outcome had clear segmentation boundary and ensured completely segmentation region. So the result from the k-mean clustering method was better. In order to reflect the advantage of high resolution image, the image processed by k-mean clustering method was contrast with Antarctic sea ice concentration data, which was received from Advanced Microwave Scanning Radiometer 2. It is fully proved that the high resolution radar data can provide the location and the size of sea ice distribution more clearly and more accurately.

The original software reliability demonstration test (SRDT) does not take adequate account of prior knowledge or the prior distribution, which can lead to an expensive use of many resources. In the current paper, we propose a new improved Bayesian based SRDT method. We begin by constructing a framework for the SRDT scheme, then we use decreasing functions to construct the prior distribution density functions for both discrete and continuous safety-critical software, and then present schemes for both discrete and continuous Bayesian software demonstration functions (which we call DBSDF and CBSDF, respectively). We have carried out a set of experiments comparing our new schemes with the classic demonstration testing scheme on several published data sets. The results reveal that the DBSDF and CBSDF schemes are both more efficient and more applicable, and this is especially the case for safety-critical software with high reliability requirements.

To merge high spatial resolution sea surface temperature (SST) in East-China Sea from multi-satellite sensors SST products, SST products from MTSAT, AVHRR, AMSR-E, MODIS and TMI in one year of 2008 were applied to analyze their availabilities and accuracies in East-China Sea. The annual-mean SST availabilities of AVHRR, AMSR-E, MODIS, TMI, and MTSAT are 22%, 33%, 40%, 56% and 61%, respectively. Compared with the relatively constant availabilities of microwave radiometers, infrared radiometers availabilities are lower and vary largely. The availability of MTSAT is the highest. The annual-mean RMSs and biases of AMSR-E, AVHRR, TMI, MODIS and MTSAT are 0.65K and 0.15K, 0.66K and 0.08K, 0.77K and 0.14K, 0.79K and -0.25K, 0.84K and -0.31K, respectively. The accuracies of AVHRR and AMSR-E are relatively better than other three sensor SSTs.

Sea surface air temperature (Ta) and relative humidity (RH) have been the main parameters of climate studies. In the past, the data can be obtained from observations, but the observations are sparse, especially over ocean. Now we can get the aid of satellites, but it is impossible to estimate them from satellites directly so far. This paper presents a new method to derive monthly averaged Ta and RH from satellite data using artificial neural networks (ANN). For Ta estimation, four inputs are needed: wind speed, cloud liquid water, total precipitable water from SSM/I and sea surface temperature (SST) from AVHRR, the data to develop and train the methodology are offered by Tropical Atmosphere Ocean (TAO) project and National Database Buoy Center (NDBC). For RH estimation, the methodology is similar with the method of Ta estimation, except adding the parameter of rain rate (from SSM/I) as the fifth inputs. Comparison with independent validation samples in the Pacific and Atlantic Oceans indicate the result of Ta and RH estimated from satellite data is reasonable well. The root mean square (RMS) and the correlation between estimated and measured air temperature are about 0.94℃ and 0.98, respectively. The RMS and the correlation of relative humidity are about 3.74 and 0.64, respectively. The simple statistical formula is also obtained in this paper. Compared with ANN methodology, the statistical formula is intuitionistic and the result is reasonable accepted.