Search

Background Experimental verification of a drug discovery process is expensive and time-consuming. Therefore, recently, the demand to more efficiently and effectively identify drug–target interactions (DTIs) has intensified. Results We treat the prediction of DTIs as a ranking problem and propose a neural network architecture, NeuRank, to address it. Also, we assume that similar drug compounds are likely to interact with similar target proteins. Thus, in our model, we add drug and target similarities, which are very effective at improving the prediction of DTIs. Then, we develop NeuRank from a point-wise to a pair-wise, and further to list-wise model. Conclusion Finally, results from extensive experiments on five public data sets (DrugBank, Enzymes, Ion Channels, G-Protein-Coupled Receptors, and Nuclear Receptors) show that, in identifying DTIs, our models achieve better performance than other state-of-the-art methods.

Collaborative filtering (CF) approaches are widely applied in recommender systems . Traditional CF approaches have high costs to train the models and cannot capture changes in user interests and item popularity. Most CF approaches assume that user interests remain unchanged throughout the whole process. However, user preferences are always evolving and the popularity of items is always changing. Additionally, in a sparse matrix , the amount of known rating data is very small. In this paper, we propose a method of online collaborative filtering with dynamic regularization (OCF-DR), that considers dynamic information and uses the neighborhood factor to track the dynamic change in online collaborative filtering (OCF). The results from experiments on the MovieLens100K, MovieLens1M, and HetRec2011 datasets show that the proposed methods are significant improvements over several baseline approaches.

As living data growing and evolving rapidly, traditional machine learning algorithms are hard to update models when dealing with new training data. When new data arrives, traditional collaborative filtering methods have to train their model from scratch. It is expensive for them to retrain a model and update their parameters. Compared with traditional collaborative filtering, the online collaborative filtering is effective to update the models instantly when new data arrives. But the cold start and data sparsity remain major problems for online collaborative filtering. In this paper, we try to utilize the convolutional neural network to extract user/item features from user/item side information to address these problems. First, we proposed a deep bias probabilistic matrix factorization (DBPMF) model by utilizing the convolutional neural network to extract latent user/item features and adding the bias into probabilistic matrix factorization to track user rating behavior and item popularity. Second, we constrain user-specific and item-specific feature vectors to further improve the performance of the DBPMF. Third, we update two models by an online learning algorithm. The extensive experiments for three datasets (MovieLens100K, MovieLens1M, and HetRec2011) show that our methods have a better performance than baseline approaches.

A model for fine-grained vehicle classification based on deep learning is proposed to handle complicated transportation scene. This model comprises of two parts, vehicle detection model and vehicle fine-grained detection and classification model. Faster R-CNN method is adopted in vehicle detection model to extract single vehicle images from an image with clutter background which may contains serval vehicles. This step provides data for the next classification model. In vehicle fine-grained classification model, an image contains only one vehicle is fed into a CNN model to produce a feature, then a joint bayesian network is used to implement the fine-grained classification process. Experiments show that vehicle’s make and model can be recognized from transportation images effectively by using our method. Furthermore,in order to build a large scale database easier, this paper comes up with a novel network collaborative annotation mechanism.

Remote medical resources configuration and management involves complex combinatorial Multi-Objective Optimization problem, whose computational complexity is a typical NP problem. Based on the MOEA/D framework, this paper applies the two-way local search strategy and the new selection strategy based on domination amount and proposes the IMOEA/D framework, following which each individual produces two individuals in mutation. In this paper, by using a new selection strategy, the parent individual is compared with two mutated offspring individuals, and the more excellent one is selected for the next generation of evolution. The proposed algorithm IMOEA/D is compared with eMOEA, MOEA/D and NSGA-II, and experimental results show that for most test functions, IMOEA/D proposed is superior to the other three algorithms in terms of convergence rate and distribution.

This paper proposed the Cloud Storage Service Selection Strategy under the cross-datacenter environment. Due to the dynamic network environment and the independence between the data centers, this paper presented Cloud Storage Service Selection Strategy across the data center based on AHP–backward cloud generator algorithm. The strategy combines the theory of analytic hierarchy process (AHP) analysis and uncertainty reasoning of cloud method by means of collecting cloud storage providers’ quantitative performance data and inferring qualitative classification of service capability, to select Cloud Storage Service Selection Strategy across the data center. Simulation results show that the strategy has a great advantage in system load balance, replica access rate, and data reliability.

We focus on the issue of dataset generation for fine-grained vehicle classification with CNN. Traditionally, to build a large dataset, images must be first collected manually, and then be annotated with a lot of effort. All these work are time-consuming and cost-prohibitive. In this work we propose a novel method that can generate massive images automatically, and these generated images need no annotation. An AutoCAD 3D model of a car of specified make and model is imported into our system, and then images of different views of the car are generated, these images can describe all the details of a car. By taking these images as training dataset, we use a Convolutional Neural Network to train a model for fine-grained vehicle classification. Experimental results show that these images generated virtually by 3D model indeed work as effective as real images.

This paper proposes a generalized cluster risk evaluation model by applying a data mining method to the cloud computing risk evaluation. The model maps data sets into a semantic space via singular value decomposition (SVD), uses a clustering algorithm to classify them and to extract the prototype vector of a particular category from clustering results, and assigns a definite weight to each category so as to set up an initial prototype vector model. The model is taken as the basis for risk evaluation of information system. After the data to be evaluated were mapped to the same semantic space, they are calculated with the prototype vector of each category, so as to obtain the similarity of the category, and the cumulative sum of the similarity with the weight of the corresponding category comes out. Finally, a mean value is calculated to obtain the risk value of the data to be evaluated, namely, the risk value of the occasion when the data is obtained. In this paper, the safety risk information is obtained from the operating system log and Web application server log of a virtual host; the Latent Semantic Analysis-based Generalized Cluster Classifier (LSA-GCC) is adopted and the MapReduce-based LSA-GCC and LSA-SAM parallel acceleration experiment is conducted. The experimental results show that in a cloud computing environment of large-scale parallel processing, the method used in this paper can identify the log events of a cloud computing system and conduct risk prompt rapidly.

The main cloud computing service providers usually provide cross-regional and services of Crossing Multi-Internet Data Centers that supported with selection strategy of service level agreement risk constraint. But the traditional quality of service (QoS)-aware Web service selection approach cannot ensure the real-time and the reliability of services selection. We proposed a cloud computing system risk assessment method based on cloud theory, and generated the five property clouds by collecting the risk value and four risk indicators from each virtual machine. The cloud backward generator integrated these five clouds into one cloud, according to the weight matrix. So the risk prediction value is transferred to the risk level quantification. Then we tested the Web service selection experiments by using risk assessment level as QoS mainly constraint and comparing with LRU and MAIS methods. The result showed that the success rate and efficiency of risk assessment with cloud focus theory Web services selection approaches are more quickly and efficient.

Since the concept of cloud computing was proposed in 2006, cloud computing has been considered as the technology that probably drives the next-generation Internet revolution and rapidly becomes the hottest topic in the field of IT. The paper synthetically introduces cloud computing techniques, including the currently non-uniform definition and the characteristics of cloud computing; The paper also introduces the core techniques of cloud computing, such as data management techniques, data storage techniques, programming model and virtualization techniques. Then the 4-tie overall technique framework of general cloud computing is talked about. Finally, the paper talks about the obstacles and opportunities.

With the continuous expansion of the cloud computing platform scale and rapid growth of users and applications, how to efficiently use system resources to improve the overall performance of cloud computing has become a crucial issue. To address this issue, this paper proposes a method that uses an analytic hierarchy process group decision (AHPGD) to evaluate the load state of server nodes. Training was carried out by using a hybrid hierarchical genetic algorithm (HHGA) for optimizing a radial basis function neural network (RBFNN). The AHPGD makes the aggregative indicator of virtual machines in cloud, and become input parameters of predicted RBFNN. Also, this paper proposes a new dynamic load balancing scheduling algorithm combined with a weighted round-robin algorithm, which uses the predictive periodical load value of nodes based on AHPPGD and RBFNN optimized by HHGA, then calculates the corresponding weight values of nodes and makes constant updates. Meanwhile, it keeps the advantages and avoids the shortcomings of static weighted round-robin algorithm.

Job scheduling technology is one of the Hadoop platform’s key technologies, whose main function is to control the execute sequence of job and the distribution of computing resources, which directly relates to the Hadoop platform’s overall performance and system resources, usage. However, the existing job scheduling algorithms such as FIFO Scheduler, Fair Scheduler and Capacity Scheduler all have some defects. To overcome theses defects, this paper proposed a new algorithm Hadoop Greedy Scheduler Based on the Fair (HGSF). Firstly, the job pools are sorted by priority from high to low; pools in the same priority are sorted by their minimum requirements. Then their minimum requirements are met in turn. If have spare computing resources, it will be assigned to the job pool which has the highest priority and the minimum difference between the ideal requirement and the minimum requirement. Finally, the algorithm uses delay allocation strategy to improve the localization of the data for computing tasks. The experimental results testified the effectiveness of the proposed algorithm.

It has been the academic researching subject how to objectively compare the various face identification algorithms. This paper design a system based on the Colorado State University identification evaluation system, which can compare face identification algorithms under application-specific. First of all, introduced the old system; then taken the logging system for desktop of PC as an example, designed the comparing system of face identification algorithms under application-specific, coded the process of achieving and pre-processed data and described the steps of processing data; at last tested PCA_MahCosine, LDA_ldaSoft, Bayesian_ML, PCA_Euclidean and Bayesian_MAP five common face identification methods, conclude Bayesian_ML algorithm is the best algorithm in face recognition logging system for logging on PC desktop and give the parameter in the system, provide a sample for testing face identification algorithms under application-specific.

Intelligent agent-based systems are regarded as the promising technology in bridging the gap between the physical world and cyber-applications. In spite of the rising demands for reusable information systems; current designs are still insufficient in providing efficient reusable mechanisms for system design. One of the major problems hinders the development of information reuse in most traditional systems is the lack of the autonomous character among system modules or subsystems. The emergence of agent technology is able to solve the problem plaguing many traditional systems. Existing agent design models create an agent as a sole system with built-in domain-specific capabilities. However, this design pattern causes several problems while matching and updating agents’ capabilities due to the built-in design pattern in these models decreases agents’ extensibility, flexibility and reusability. In this paper we introduce a novel design for agent-based systems, which is able to provide an efficient design pattern for improving the reusability, extensibility and flexibility of agent design. The novel agent capability design offers an open and flexible structure; and implements several practical algorithms that can improve the system performance. An experimental program based on several practical cases has been developed to evaluate the performance of the proposed design. The empirical results reveal the efficiency of the new agent design pattern.

Credit scoring is becoming a competitive issue with rapid growth and significant advance. Building a satisfactory credit model has attracted lots of researchers in the past decades and it is still one of the hottest research topics in the field of credit industry. This paper emphasizes on surveying the development of the artificial intelligence technologies for solving credit scoring. It covers algorithms including the support vector machines, artificial neural networks, genetic algorithms, genetic programming algorithms and their hybridization.

Conference Name:2013 International Conference on Future Computer and Information Technology, ICFCIT 2013. Conference Address: Tianjin, China. Time:August 22, 2013 - August 23, 2013.

With the rapid update of internet hardware and software, especially the rise of mobile internet and cloud computing platforms, web-based sites has become the world's largest source of information, and so how to extract the accurate domain-specific knowledge, how to improve the extraction efficiency and how to save the cost during information harvest have become the hot research spot. With the advantage of the theories and techniques of software architecture, the paper presents a web knowledge discovery model based on SOA and cloud computing, through which valid information can be obtained in an automatic or semi-automatic way from the distributed information sources environment. A unique web information extraction system built base on the architecture and the verification results were given in the final chapter.

Cloud computing is usually used to describe the large-scale distributed infrastructure, platform and software services provided by third party, which is a hot topic in IT area in recent year. However, it is disputable in industrial and academic area that what is cloud computing and whether it is a novel technology. This paper attempts to draw a clear picture by the introduction of several public commercial clouds and open source cloud computing software, not to discuss the exactly concept of cloud computing. The commercial cloud computing Amazon EC2 IBM smart cloud Google App Engine and Windows Azure were introduced and the implementation of several open source cloud computing software such as Hadoop Eucalyptus open Nebula and Nimbus were detailed. At last, several representative definitions were showed. The purpose of this paper is that a basic knowledge about cloud computing can be obtained by reading this paper.

The grid computational environment is very suit to meet the computational demands of large, diverse groups of tasks. And the task scheduling problem in it has been a research hotspot in recent years. Some heuristic methods have been used to optimize it and have got some good results. However, selecting the best one to use in a given environment remains a difficult problem, because comparisons are often clouded by different underlying assumptions in the original study of each algorithm. Some comparisons have been made to them, but some new algorithms are not included in the comparisons. So, in this paper, four popular researched algorithms recently are selected, implemented, and analyzed. The four heuristics are Genetic Algorithm, Ant Colony Optimization Algorithm, Particle Swarm Optimization Algorithm and Simulated Annealing Algorithm. The evaluations include the schedule creating time, the makespan and the mean response time. It shows that for the cases studied here, the PSO heuristic performs better in comparison to the other techniques.

Grid Workflow Scheduling represented by DAG(Directed Acyclic Graph) is a typical NP-complete problem, and thus a scheduling algorithm of high efficiency is required. So an improved genetic algorithm is proposed to solve this problem. In the algorithm, chromosomes of poor fitness make secondary preferential hybridization and mutation with the overall best individual. It not only guarantees the population diversity but increases the convergence rate of population. Experiment results based on Gridsim prove it available and better than standard genetic algorithm.

Audio effects processing technology is widely used in various types of music production, live performances and other occasions. Most of traditional technologies produce real-time audio effects with hardware analog circuits. In this paper, focus on the demand of time scaling, pitch shifting and special effects of real-time audio. Improve the algorithm performance, efficiency and effectiveness in subjective base on traditional phase vocoder model in high performance DSP processor BlackFin which be made the use of its structural features. Experiment shows that, new algorithm do well in both computation and hearing effect.