International Journal on Artificial Intelligence Tools

An automated algorithm is presented to determine the DNA molecule intrinsic curvature profiles and the molecular spatial orientations in Atomic Force Microscope images. The curvature is composed by static and dynamic contributions. The former is the intrinsic curvature, a function of the DNA nucleotide sequence, while the latter is due to thermal fluctuations. This algorithm allows to reconstruct the intrinsic curvature profile excluding the thermal contribution. The DNA intrinsic curvature profile is computed in consequence of the detection of the correct spatial orientation of the molecules on the AFM substrate following the DNA deposition process. To discover the correct molecular orientations, we propose a fast heuristic orientation finding algorithm, that modifies one DNA molecular orientation at a time with linear-time heuristic transitions.

We propose a model selection method to estimate the relation of multiple SNPs, environmental factors and the binary disease trait. We applied the combination of logistic regression and genetic algorithm for this study. The logistic regression model can capture the continuous effects of environments without categorization, which causes the loss of the information. To construct an accurate prediction rule for binary trait, we adopted Akaike's information criterion (AIC) to find the most effective set of SNPs and environments. That is, the set of SNPs and environments that gives the smallest AIC is chosen as the optimal set. Since the number of combinations of SNPs and environments is usually huge, we propose the use of the genetic algorithm for choosing the optimal SNPs and environments in the sense of AIC. We show the effectiveness of the proposed method through the analysis of the case/control populations of diabetes, Alzheimer's disease and obesity patients. We succeeded in finding an efficient set to predict types of diabetes and some SNPs which have strong interactions to age while it is not significant as a single locus.

Predicting native conformations using computational protein models requires a large number of energy evaluations even with simplified models such as hydrophobic-hydrophilic (HP) models. Clearly, energy evaluations constitute a significant portion of computational time. We hypothesize that given the structured nature of algorithms that search for candidate conformations such as stochastic methods, energy evaluation computations can be cached and reused, thus saving computational time and effort. In this paper, we present a caching approach and apply it to 2D triangular HP lattice model. We provide theoretical analysis and prediction of the expected savings from caching as applied this model. We conduct experiments using a sophisticated evolutionary algorithm that contains elements of local search, memetic algorithms, diversity replacement, etc. in order to verify our hypothesis and demonstrate a significant level of savings in computational effort and time that caching can provide.

After introduction to medical imaging, virtual reality (VR), and VR surgical simulation, the paper overviews our long-term efforts in human body modeling and development of VR-based applications for brain intervention. It describes construction, features, and applications of anatomical, functional, and vascular brain atlases. It also presents VR-based applications developed for stereotactic and functional neurosurgery (NeuroPlanner, NeuroBase, and BrainBench), interventional neuroradiology (NeuroCath and tele-NeuroCath), and tumor stereotaxis (VIVIAN). Finally, a vision of the future surgery (the DOTELL environment) is presented.

Several of ten thousands functional genes control the growth, genetics, and behavior of living organisms by regulating different gene expressions. The genes in a normal cell control the process of cell growth, differentiation, reproduction, and apoptosis via multiple steps of interactive regulation mechanism. The mechanism of gene regulation is a very important process in human beings. If there is something wrong in the gene regulation mechanism, it may cause some diseases such as cancer. It is very difficult to identify the regulatory relations among genes in human genome. Traditional biological research methods consume huge amount of time and man strength to do this work. In recent years, with the rapid development of microarray technologies, cDNA can be used to analyze the changes of gene expressions in different cells in a high throughput manner. In this paper, we propose a novel bioinformatics approach to predict possible cancerous genes based on a so-called regulation truth table (RTT) of genes. The RTT of two genes is constructed using the differential expressions of cDNA microarray data for tumor and normal tissues. The differences in regulatory relations of genes for tumor and normal tissues are adopted to identify possible cancerous genes.

AutoTutor is an intelligent tutoring system that holds conversations with learners in natural language. AutoTutor uses Latent Semantic Analysis (LSA) to match student answers to a set of expected answers that would appear in a complete and correct response or which reflect common but incorrect understandings of the material. The correctness of student contributions is decided using a threshold value of the LSA cosine between the student answer and the expectations. In previous work LSA has shown to be effective in detecting good answers of students. The results indicate that the best agreement between LSA matches and the evaluations of subject matter experts can be obtained if the cosine threshold is allowed to be a function of the lengths of both student answer and the expectation being considered. Based on some of our experiences with LSA and AutoTutor, we are developing a new mathematical model to improve the precision of AutoTutor's natural language understanding and discriminative ability.

Recent advances in solving constraint satisfaction problems (CSPs) and heuristic search have made it possible to solve classical planning problems significantly faster. There is an increasing amount of work on extending these advances to solving planning problems in more expressive languages. These problems and languages contain metric time, quantifiers and resource quantities. SAT-based approaches are very effective at optimal planning. In this paper we report on SAT-based temporal planner T-SATPLAN. One key challenge in the development of planners casting planning as SAT or CSP is the identification of constraints which are satisfied if and only if there is a plan of k steps. In this paper we show how such a SAT encoding can be synthesized for temporal planning. As part of this, we generalize explanatory frame axioms for two states of fluents (true, false) to three states of fluents (true, false and undefined). We show how this SAT encoding can be simplified. We discuss two additional SAT encodings of temporal planning. The encoding schemes make it easier to exploit progress in SAT and CSP solving to solve temporal planning problems. We also report on an experimental evaluation of T-SATPLAN using one such encoding scheme. The evaluation shows that significantly large SAT encodings of temporal planning problems can be solved extremely fast.

In this paper, the route-planning problems of Unmanned Aerial Vehicle (UAV) in uncertain and adversarial environment are addressed, including not only single-mission route planning in known a priori environment, but also the route replanning in partially known and mission-changeable environments. A mission-adaptable hybrid route-planning algorithm based on flight roadmap is proposed, which combines existing global and local methods (Dijkstra algorithm, SAS and D*) into a two-level framework. The environment information and constraints for UAV are integrated into the procedure of building flight roadmap and searching for routes. The route-planning algorithm utilizes domain-specific knowledge and operates in real time with near-optimal solution quality, which is important to uncertain and adversarial environment. Other planners do not provide all of the functionality, namely real-time planning and replanning, near-optimal solution quality, and the ability to model complex 3D constraints.

Authorship attribution can assist the criminal investigation procedure as well as cybercrime analysis. This task can be viewed as a single-label multi-class text categorization problem. Given that the style of a text can be represented as mere word frequencies selected in a language-independent method, suitable machine learning techniques able to deal with high dimensional feature spaces and sparse data can be directly applied to solve this problem. This paper focuses on classifier ensembles based on feature set subspacing. It is shown that an effective ensemble can be constructed using, exhaustive disjoint subspacing, a simple method producing many poor but diverse base classifiers. The simple model can be enhanced by a variation of the technique of cross-validated committees applied to the feature set. Experiments on two benchmark text corpora demonstrate the effectiveness of the presented method improving previously reported results and compare it to support vector machines, an alternative suitable machine learning approach to authorship attribution.

The obvious need for using modern computer networking capabilities to enable the effective sharing of information has resulted in data-sharing systems, which store, and manage large amounts of data. These data need to be effectively searched and analyzed. More specifically, in the presence of dirty data, a search for specific information by a standard query (e.g., search for a name that is misspelled or mistyped) does not return all needed information, as required in homeland security, criminology, and medical applications, amongst others. Different techniques, such as soundex, phonix, n-grams, edit-distance, have been used to improve the matching rate in these name-matching applications. These techniques have demonstrated varying levels of success, but there is a pressing need for name matching approaches that provide high levels of accuracy in matching names, while at the same time maintaining low computational complexity. In this paper, such a technique, called ANSWER, is proposed and its characteristics are discussed. Our results demonstrate that ANSWER possesses high accuracy, as well as high speed and is superior to other techniques of retrieving fuzzy name matches in large databases.

Security in web-based systems that handle confidential information can be considered a particularly sensitive subject that requires assuming some responsibilities about security. Achieving a secure web application involves tackling several issues such encryption of traffic and certain database information, strictly restricted access control, etc. In this work we focus on detecting misuse of the web application in order to gain unauthorised access. We introduce an Intrusion Detection component that by applying Text Categorisation is capable of learning the characteristics of both normal and malicious user behaviour from the regular, high-level log entries generated by web application through its application server. Therefore, the detection of misuse in the web application is achieved without the need of explicit programming or modification of the existing web application. We applied our Intrusion Detection component to a real web-based telemedicine system in order to offer some evaluation measurements. This articles offers an overview of the model, our experiences, and observations.

Reinforcement learning (RL) has been successfully used in many fields. With the increasing complexity of environments and tasks, it is difficult for a single learning algorithm to cope with complicated problems with high performance. This paper proposes a new multiple learning architecture, "Aggregated Multiple Reinforcement Learning System (AMRLS)", which aggregates different RL algorithms in each learning step to make more appropriate sequential decisions than those made by individual learning algorithms. This architecture was tested on a Cart-Pole system. The presented simulation results confirm our prediction and reveal that aggregation not only provides robustness and fault tolerance ability, but also produces more smooth learning curves and needs fewer learning steps than individual learning algorithms.