Advances in Computational Intelligence Systems

Association rule mining can be combined with complex network theory to automatically create a knowledge base that reveals how certain drugs cause side-effects on patients when they interact with other drugs taken by the patient when they have two or more diseases. The drugs will interact with on-target and off-target proteins often in an unpredictable way. A computational approach is necessary to be able to unravel the complex relationships between disease comorbidities. We built statistical models from the publicly available FAERS dataset to reveal interesting and potentially harmful drug combinations based on side-effects and relationships between co-morbid diseases. This information is very useful to medical practitioners to tailor patient prescriptions for optimal therapy.

Since the long short-term memory (LSTM) network is a sequential structure, it is difficult to effectively represent the structural level information of the context. Sentiment analysis based on the original LSTM causes a problem of structural level information loss, and its capacity to capture the context information is finite. To address this problem, we proposed a novel structure-attention-based LSTM as a hierarchical structure model. It may capture relevant information in the context as much as possible. We propose HM (h_t matrix) to storage the structural information of the context. Furthermore, we introduce the attention mechanism to realize vector selection. Compared with the original LSTM and normal attention-based sentiment classification methods, our model obtains a higher classification precision. It is proved that the structure-attention-based method proposed in this study has an advantage in capturing the potential semantic structure.

The need for adaptive systems is growing with the increasing number of autonomous entities such as software systems and robots. A key characteristic of adaptive systems is that their environment changes, possibly in ways that were not envisaged at design-time. These changes in requirements, model and context mean the functional behaviour of a system cannot be fully defined in many cases, and consequently formal verification of the system is not possible. In this research, we propose a fuzzy representation to describe the result of requirement verification. We use an adaptive assisted living system as the case study. The RELAX language is used to create a flexible system specification. We model and simulate the system using UPPAAL 4 and use a fuzzy approach to translate the simulation result into fuzzy requirement satisfaction. The result shows the benefit of a more flexible representation by describing the degree of requirement satisfaction rather than a strict yes/no Boolean judgment.

Measuring semantic similarity is essential to many natural language processing (NLP) tasks. One widely used method to evaluate the similarity calculating models is to test their consistency with humans using human-scored gold-standard datasets, which consist of word pairs with corresponding similarity scores judged by human subjects. However, the descriptions on how such datasets are constructed are often not sufficient previously. Many problems, e.g. how the word pairs are selected, whether or not the scores are reasonable, etc., are not clearly addressed. In this paper, we proposed a multidisciplinary method for building and validating semantic similarity standard datasets, which is composed of 3 steps. Firstly, word pairs are selected based on computational linguistic resources. Secondly, similarities for the selected word pairs are scored by human subjects. Finally, Event-Related Potentials (ERPs) experiments are conducted to test the soundness of the constructed dataset. Using the proposed method, we finally constructed a Chinese gold-standard word similarity dataset with 260 word pairs and validated its soundness via ERP experiments. Although the paper only focused on constructing Chinese standard dataset, the proposed method is applicable to other languages.

Identifying relationship between attribute variables from different data sources is an emerging field in data mining. However, currently there seldom exist effective methods designed for this particular problem. In this paper, a novel approach for inter-variable correlation prediction is proposed through the employment of the concept of connected-triple, and implemented with fuzzy logic. By the use of link strength measurements and fuzzy inference, the job of detecting similar or related variables can be accomplished via examining the link relation patterns. Comparative experimental investigations are carried out, demonstrating the potential of the proposed work in generating acceptable predicted results, while involving only simple computations.

Anatomical Therapeutic Codes (ATC) are a drug classification system which is extensively used in the field of drug development research. There are many drugs and medical compounds that as yet do not have ATC codes, it would be useful to have codes automatically assigned to them by computational methods. Our initial work involved building feedforward multi-layer perceptron models (MLP) but the classification accuracy was poor. To gain insights into the problem we used the Kohonen self-organizing neural network to visualize the relationship between the class labels and the independent variables. The information gained from the learned internal clusters gave a deeper insight into the mapping process. The ability to accurately predict ATC codes was unbalanced due to over and under representation of some ATC classes. Further difficulties arise because many drugs have several, quite different ATC codes because they have many therapeutic uses. We used chemical fingerprint data representing a drugs chemical structure and chemical activity variables. Evaluation metrics were computed, analysing the predictive performance of various self-organizing models.

Animal social networks offer an important research mechanism for animal behaviour analysis. Inferring social network structures in ecological systems from spatiotemporal data streams [1] presents a procedure to build such networks based on animal’s foraging process data which consists of time and location records. The method clusters the individuals into different gathering events and links up the individuals that appear in the same events, and subsequently filters coincident links. However, the original model does not perform well in many aspects, including time and space complexity and not-unique coincident link filtering threshold. To modify this method, fuzzy c-means is employed in this work to cluster all links into two groups, strong links or weak links. The work presented here also experimentally compares the performance of the proposed modification against the original method, demonstrating the efficacy of the modified version.

This study considers the dynamic frequency assignment problem, where new requests gradually become known and frequencies need to be assigned to those requests effectively and promptly with the minimum number of reassignments. The problem can be viewed as a combination of three underlying problems: the initial problem, the online problem, and the repair problem. In this study, a heuristic approach is proposed to solve this problem using different solution methods for each underlying problem. Moreover, the efficiency of this approach is improved by means of the Gap technique, which aims to identify a good frequency to be assigned to a given request. For the purpose of this study, new dynamic datasets are generated from static benchmark datasets. It was found that the performance of our approach is better than the state-of-the-art approach in the literature across the same set of instances.

Ant Colony Optimisation (ACO) is a well known metaheuristic that has proven successful at solving Travelling Salesman Problems (TSP). However, ACO suffers from two issues; the first is that the technique has significant memory requirements for storing pheromone levels on edges between cities and second, the iterative probabilistic nature of choosing which city to visit next at every step is computationally expensive. This restricts ACO from solving larger TSP instances. This paper will present a methodology for deploying ACO on larger TSP instances by removing the high memory requirements, exploiting parallel CPU hardware and introducing a significant efficiency saving measure. The approach results in greater accuracy and speed. This enables the proposed ACO approach to tackle TSP instances of up to 200K cities within reasonable timescales using a single CPU. Speedups of as much as 1200 fold are achieved by the technique.

The classical algorithms based on regularization usually solve sparse optimization problems under the framework of single objective optimization, which combines the sparse term with the loss term. The majority of these algorithms suffer from the setting of regularization parameter or its estimation. To overcome this weakness, the extension of multiobjective evolutionary algorithm based on decomposition (MOEA/D) has been studied for sparse optimization. The major advantages of MOEA/D lie in two aspects: (1) free setting of regularization parameter and (2) detection of true sparsity. Due to the generational mode of MOEA/D, its efficiency for searching the knee region of the Pareto front is not very satisfactory. In this paper, we proposed a new steady-state MOEA/D with the preference to search the region of Pareto front near the true sparse solution. Within each iteration of our proposed algorithm, a local search step is performed to examine a number of solutions with similar sparsity levels in a neighborhood. Our experimental results have shown that the new MOEA/D clearly performs better than its previous version on reconstructing artificial sparse signals.

In this paper, we present a method to generate map sketches for strategy games using a state of the art many-objective evolutionary algorithm, namely NSGAIII. The map sketch generator proposed in this study outputs a three objective Pareto-front in which all the points are fair and strong in different aspects. The generated map sketch can be used by level designers to create real time strategy maps effectively and/or help them see multiple aspects of a game map simultaneously. The algorithm can also be utilised as a benchmark generator to be used in tests for various cases such as shortest path algorithms and strategy game bots. The results reported in this paper are very promising and promote further study.

In this paper, we propose a reference-inspired multiobjective evolutionary algorithm for many-objective optimisation. The main idea is (1) to summarise information inspired by a set of randomly generated reference points in the objective space to strengthen the selection pressure towards the Pareto front; and (2) to decompose the objective space into subregions for diversity management and offspring recombination. We showed that the mutual relationship between the objective vectors and the reference points provides not only a fine selection pressure, but also a balanced convergence-diversity information. The decomposition of the objective space into subregions is able to preserve the Pareto front’s diversity. A restricted stable match strategy is proposed to choose appropriate parent solutions from solution sets constructed at the subregions for high-quality offspring generation. Controlled experiments conducted on a commonly-used benchmark test suite have shown the effectiveness and competitiveness of the proposed algorithm in comparison with several state-of-the-art many-objective evolutionary algorithms.

Dimension reduction of data is an important issue in the data processing and it is needed for the analysis of higher dimensional data in the application domain. Reduct in the rough set is a minimal subset of features, which has the same discernible power as the entire features in the higher dimensional scheme. In this paper, generations of reducts and threshold functions are developed for the classification system. The reduct followed by the nearest neighbor method or threshold functions is useful for the reduct classification system. For the classification, a nearest neighbor relation with minimal distance proposed here has a fundamental information for classification. Then, the nearest neighbor relation plays a fundamental role on the discernibility and in discernibility matrices, in which the indiscernibility matrix is proposed here to test the sufficient condition for reduct and threshold function. Then, generation methods for the reducts and threshold functions based on the nearest neighbor relation are proposed here using Boolean operations on the discernibility and the indiscernibility matrices.

This paper proposes a fuzzy brain emotional learning network for adaptive noise cancelation. The proposed network is based on brain emotional learning algorithm which is developed according to the emotional learning process of mammalian and the fuzzy inference is added for better ability to handle uncertainties. Parameters in the network are modified online by the derived adaption laws. In addition, a stable convergence is guaranteed by utilizing the Lyapunov stability theorem. Finally, in order to demonstrate the performance of the proposed filter, it is applied in a signal processing application where different source signals and noise signals are used. A comparison between the proposed method, Least mean square algorithm and a fuzzy cerebellar model articulation controller filter shows that the proposed method can converge faster even when the source signal is corrupted severely.

Lung cancer is a widespread disease and it is well understood that systematic, non-invasive and early detection of this progressive and life-threatening disorder is of vital importance for patient outcomes. In this work we present a convergence of familiar and less familiar artificial neural network techniques to help address this task. Our preliminary results demonstrate that improved, automated, early diagnosis of lung cancer based on the classification of volatile organic compounds detected in the exhaled gases of patients seems possible. Under strictly controlled conditions, using Selected Ion Flow Tube Mass Spectrometry (SIFT-MS), the naturally occurring concentrations of a range of volatile organic compounds in the exhaled gases of 20 lung cancer patients and 20 healthy individuals provided the dataset that has been analysed. We investigated the performance of several artificial neural network architectures, each with complementary pattern recognition properties, from the domains of supervised, unsupervised and recurrent neural networks. The neural networks were trained on a subset of the data, with their performance evaluated using unseen test data and classification accuracies ranging from 56% to 74% were obtained. In addition, there is promise that the topological ordering properties of the unsupervised networks’ clusters will be able to provide further diagnostic insights, for example into patients who may have been heavy smokers but so far have not presented with any lung cancer. With the collection of data from a larger number of subjects across a long time period there is promise that an automated assistive tool in the diagnosis of lung cancer via breath analysis could soon be possible.

The ability to predict future states is fundamental for a wide variety of applications, from weather forecasting to stock market analysis. Understanding the related data attributes that can influence changes in time series is a challenging task that is critical for making accurate predictions. One particular application of key interest is understanding the factors that relate to the occurrence of global activities from online world news reports. Being able to understand why particular types of events may occur, such as violence and peace, could play a vital role in better protecting and understanding our global society. In this work, we explore the concept of predicting the occurrence of world news events, making use of Global Database of Events, Language and Tone online news aggregation source. We compare traditional Auto-Regressive Moving Average models with more recent deep learning strategies using Long Short-Term Memory Recurrent Neural Networks. Our results show that the latter are capable of achieving lower error rates. We also discuss how deep learning methods such as Recurrent Neural Networks have the potential for greater capability to incorporate complex associations of data attributes that may impact the occurrence of future events.

AES, ElGamal are two examples of algorithms that have been developed in cryptography to protect data in a variety of domains including native and cloud systems, and mobile applications. There has been a good deal of research into the use of side channel attacks on these algorithms. This work has conducted an experiment to detect malicious loops inside Flush+Reload and Prime+Prob attack programs against AES through the exploitation of Hardware Performance Counters (HPC). This paper examines the accuracy and efficiency of three machine learning algorithms: Neural Network (NN); Decision Tree C4.5; and K Nearest Neighbours (KNN). The study also shows how Standard Performance Evaluation Corporation (SPEC) CPU2006 benchmarks impact predictions.

Deciphering user intent from website clickstreams and providing more relevant product recommendations to users remains an important challenge in Ecommerce. We outline our approach to the twin tasks of user classification and content ranking in an Ecommerce setting using an open dataset. Design and development lessons learned through the use of gradient boosted machines are described and initial findings reviewed. We describe a novel application of word embeddings to the dataset chosen to model item-item similarity. A roadmap is proposed outlining future planned work.

Keyphrases provide important semantic metadata for organizing and managing free-text documents. As data grow exponentially, there is a pressing demand for automatic and efficient keyphrase extraction methods. We introduce in this paper SemCluster, a clustering-based unsupervised keyphrase extraction method. By integrating an internal ontology (i.e., WordNet) with external knowledge sources, SemCluster identifies and extracts semantically important terms from a given document, clusters the terms, and, using the clustering results as heuristics, identifies the most representative phrases and singles them out as keyphrases. SemCluster is evaluated against two baseline unsupervised methods, TextRank and KeyCluster, over the Inspec dataset under an F1-measure metric. The evaluation results clearly show that SemCluster outperforms both methods.

P300-based brain-computer interface (BCI) has been widely studied over two decades. However, there are several factors that hamper P300-based BCI to be used in daily life. EEG acquisition devices are often too much expensive for an average customer. Although the Emotiv Epoc headset is a kind of low-cost device for recording brain signals and has been adopted to develop some BCI systems, due to the limited number of electrodes, the Emotiv Epoc headset cannot cover the regions of scalp that are convenient for detecting P300, so the effectiveness of the Emotiv Epoc headset used in the P300-based BCI has been doubted by many researchers. This paper aims to examine the performance of Emotiv Epoc headset used in the P300-based BCI system. Six participants participated in the experiment and two paradigms were compared. The results demonstrated that P300 could be effectively detected from the brain signals recorded by the Emotiv Epoc headset, showing the promising future to develop low-cost P300-based BCI systems.

It is very useful for the E-learning systems to detect the students emotional state accurately, and this can remind the teacher in time to change the teaching rhythm or content to meet the student’s emotional changes for making the teaching effect optimization. In this paper, we propose an emotion detection method based on a deep learning approach, Expectation-maximization Deep Spatial-Temporal Inference Network (EM-DeSTIN). This method takes the student’s facial expression as input and combine with Support Vector Machine (SVM) to implement emotion classification and identification. Experimental results show that the proposed method improves the performance of detecting emotion in a noisy environment compared with other methods.

Assisted living homes aim to deploy tools to promote better living of elderly population. One of such tools is assistive robotics to perform tasks a human carer would normally be required to perform. For assistive robots to perform activities without explicit programming, a major requirement is learning and classifying activities while it observes a human carry out the activities. This work proposes a human activity learning and classification system from features obtained using 3D RGB-D data. Different classifiers are explored in this approach and the system is evaluated on a publicly available data set, showing promising results which is capable of improving assistive robots performance in living environments.

Cerebral Visual Impairment (CVI) is a medical area that concerns the study of the effect of brain damages on the visual field (VF). People with CVI have difficulties in their mobility and they have behaviours that others find hard to understand due to their visual impairment. A branch of Artificial Intelligence (AI) is the simulation of behaviour by building computational models that help to explain how people solve problems or why they behave in a certain way. This paper describes a novel computational system that simulates the navigation problem that is faced by people with CVI. This will help relatives, friends, and ophthalmologists of CVI patients understand more about their difficulties in navigating their everyday environment.The navigation simulation system is implemented using the Unity3D game engine. Virtual scenes of different living environment are also created using the Unity modelling software. The vision of the avatar in the virtual environment is implemented using a camera provided by the 3D game engine. Filters that mimic visual defects are created automatically and placed in front of the visual field of the avatar. The filters are based on the visual field charts of individual patients. Algorithms for navigation based on the limited vision have also been developed to demonstrate navigation problems because of the visual defects. The results showed different actions for the navigation behaviours according to the patients’ vision, and the navigations differ from patient to another according to their different defects.

Physically falling down amongst the elder helpless party is one of the most intractable issues in the era of ageing society, which has attracted intensive attentions in academia ranging from clinical research to computer vision studies. This paper proposes a fall detection/recognition system within the realm of computer vision. The proposed system integrates a group of gradient-based local visual feature extraction approaches, including histogram of oriented gradients (HOG), histogram of motion gradients (HMG), histogram of optical flow (HOF), and motion boundary histograms (MBH). A comparative study of the descriptors with the support of an artificial neural network was conducted based on an in-house captured dataset. The experimental results demonstrated the effectiveness of the proposed system and the power of these descriptors in real-world applications.

In this paper we propose a computational methodology for assessing the impact of trust associated to sources of information in situational understanding activities—i.e. relating relevant information and form logical conclusions, as well as identifying gaps in information in order to answer a given query. Often trust in the source of information serves as a proxy for evaluating the quality of the information itself, especially in the cases of information overhead. We show how our computational methodology support human analysts in situational understanding by drawing conclusions from defaults, as well as highlighting issues that demand further investigation.

Analysis of traffic condition is of great significance to urban planning and public administration. However, traditional traffic condition analysis approaches mainly rely on sensors, which are high-cost and limit their coverage. To solve these problems, we propose a semi-supervised learning method which uses the social network data instead and analyzes the traffic condition based on user’s emotion tendency. First we train the Gated Recurrent Unit (GRU) model to estimate the sentiment of microblog with traffic information, then using the emotional tendency to predict whether traffic jams happen or not. In order to reduce the data annotated by manpower, we propose a new idea to employ the Conditional Generative Adversarial Networks (CGAN) to generate samples which are as a supplement to the training set of GRU. Finally compared with the GRU model trained by solely the manual annotation data, our method improves the classification accuracy by 4.07%. We also use our model to predict the time and roads of traffic jams in 4 Chinese cities which is proved to be effective.

Timetable optimization is essential to the improvement of a bus operating company’s economic profits, quality of service and competitiveness in the market. The most previous researches studied the bus timetabling with assuming the passenger demand is certain but it varies in practice. In this study, we consider a timetable optimization problem of a single bus line under fuzzy environment. Assuming the passenger quantity in per time segment is a fuzzy value, a fuzzy bi-objective programming model that maximizes the total passenger volume and minimizes the total bus travel time under a capacity rate constraint is established. This chance constrained programming model is formulated with the passenger volume and capacity rate under certain chance constraints. Furthermore, a genetic algorithm of variable length is designed to solve the proposed model. Finally, we present a case study that utilizing real data obtained from a major Beijing bus operating company to illustrate the proposed model and algorithm.

This paper proposes an ant colony optimization (ACO) based algorithm to minimise the fleet size required to solve dial-a-ride problem (DARP). In this work, a static multi-vehicle case of DARP is considered where routes of multiple vehicles are designed to serve customer requests which are known a priori. DARP necessitates the need of high quality algorithm to provide optimal feasible solutions. We employ an improved ACO algorithm called ant colony system (ACS) to solve DARP. The fleet minimisation is also achieved by using ACS. In summary, multiple ACS are employed to minimise the fleet size while generating feasible solutions for DARP. Furthermore, the theoretical results are also validated through simulations.

For bus carriers, it is the most basic and important problem to create the bus scheduling timetable based on bus fleet configuration and passenger flow demand. Considering different technical and economic properties, vehicle capacities and limited available number of heterogeneous buses, as well as the time-space characteristics of passenger flow demand, this paper focuses on creating the bus timetables and sizing the buses simultaneously. A bi-objective optimization model is formulated, in which the first objective is to minimum the total operation cost, and the second objective is to maximum the passenger volume. The proposed model is a nonlinear integer programming, thus a genetic algorithm with self-crossover operation is designed to solve it. Finally, a case study in which the model is applied to a real-world case of a bus line in the city of Beijing, China, is presented.

Speedy development of the large-sized retail outlets empowers the emergence of dominant retailers, as a result of power transformation from suppliers to retailers. In this paper, we model a market comprised of a dominant entrant retailer, a weaker incumbent counterpart, and a common supplier from which both retailers source products. The retailers are quantity-competing, and the dominant retailer is entitled to determine the wholesale price it purchases, while the incumbent retailer accepts the price offered by the supplier. Besides, the incumbent retailer is assumed to hold private information about market demand. We investigate the collaboration strategy for the supplier which either cooperates with the dominant entrant retailer or with the vulnerable incumbent counterpart. Our result reveals that the supplier’s strategy depends on subtle considerations of multiple factors such as terminal market demand state, the demand fluctuation, the expected market demand and the dominant retailer’s wholesale price.

Making an allocation scheme which can achieve the optimal overall efficiency that matching multiple logistics tasks and resources under the environment that the tasks’ demands are uncertain is difficult. In this paper, we build a mathematical model to describe the problem and try to solve it by the genetic algorithm. We also consider the daily usage amount of each resource should be as equilibrious as possible. The result of the case simulation proves the effectiveness of the model and the algorithm. As well as, we analyze the impact that the size of the uncertainty’s degree on the allocation result.

Vehicle routing problem (VRP) plays a vital role in logistics management. Among which, the transportation of hazardous material attracts much attention especially in China. The hazardous material multi-depot vehicle routing problem (HMDVRP) considers the transportation of hazardous material and multiple depots based on VRP. This paper develops a new HMDVRP bi-objective optimization model. Some new decision variables are introduced to the model to describe the sequence of customers and simplify the model expression. Moreover, the risk measurement of the model considers the change of the loading, which reflects the nature of hazardous material transportation. HMDVRP is NP-hard, and the heuristic algorithms are the main method used for solving it. This paper proposes a two-stage heuristic algorithm to solve the new HMDVRP model. Numerical experiments show that the two-stage heuristic algorithm can solve the HMDVRP model effectively and efficiently.