International Journal on Artificial Intelligence Tools

Packing and cutting problems arise in a wide variety of industrial situations. The basic problem is that of determining a good arrangement of objects in a region without any overlap. Much research has been done on two and three dimensional rectangular packing while there has been little work done on irregular packing. In this work, we study the two-dimensional irregular packing problem and provide heuristic solutions which use rectilinear and piecewise-linear representations of objects. These heuristics include Genetic Algorithms and Tabu Search. Experimentation gives good results.

A method has been developed and implemented that assigns syntactic roles to commas. Text that has been tagged using a part-of-speech tagger serves as the input to the system. A set of Finite State Automata first assigns temporary syntactic roles to each comma in the sentence. A greedy learning algorithm is then used to determine the final syntactic roles of the commas. The system requires no training and is not domain specific. The performance of the system on numerous corpora is given and compared against a rule-based approach.

Planning with temporally extended goals has recently been the focus of much attention to researchers in the planning community. We study a class of planning goals where in addition to a main goal there exist other goals, which we call auxiliary goals, that act as constraints to the main goal. Both these type of goals can, in general, be a temporally extended goal. Linear temporal logic (LTL) is inadequate for specification of the overall goals of this type, although, for some situations, it is capable of expressing them separately. A branching-time temporal logic, like CTL, on the other hand, can be used for specifying these goals. However, we are interested in situations where an auxiliary goal has to be satisfiable within a fixed bound. We show that CTL becomes inadequate for capturing these situations. We bring out an existing logic, called min-max CTL, and show how it can effectively be used for the planning purpose. We give a logical framework for expressing the overall planning goals. We propose a sound and complete planning procedure that incorporates a model checking technology. Doing so, we can answer such planning queries as plan existence at the onset besides producing an optimal plan (if any) in polynomial time.

Service functions are major component of every industry. The evolution of service functions have seen manually intensive tasks being replaced by a combination of mental and interactive tasks. As a part of this transformation, we have seen an increasing dependency on the high technology, actually information technology, mostly with computer-communications and the web. This high technology will finally replace the service functions by a combination of knowledge intensive, interactive and customer's individual tasks in the future. However, these knowledge intensive, interactive and customer's individual service functions could be enjoyable for the restricted person, due to the difficulties of the man (customers)-machine (PCs, web-server machines) interface and also the service providers are not able to obtain much more customer's individual and abstruse information for providing the customer's individual services, due to the customers' privacy problem. In this paper we focus on of knowledge intensive, interactive and customer's individual services in the future. We emphasize the importance of developing the mechanism of humanization for making these services enjoyable for anyone and the personalization for enjoyable the more individual services, and show some examples for these mechanisms in which humanization is realized by humanoid robots and personalization is realized by customers' privacy preserving scheme. We present a feedback system including the mechanism for humanization and personalization. By performing this feedback system, anyone who is not familiar to machines, can interact with machines like human-beings and can poll or report even though privacy related information anonymously to service providers. By periodically performing this feedback system, the customers' knowledge, information, suggestions and ideas can could be accumulated and the service providers could provide more individual services to customers, Once humanization and personalization mechanism are developed, by performing our feedback systems, we enhance customers' convenience and reduce the incidence of operation errors and thereby improve their productivity and quality of their life.

We address in this paper a problem of autonomous agents performing in a common environment where each agent has a goal to achieve before a given deadline. Each agent must determine a local plan and then react properly to the local plans of other agents before a given deadline. The solution presented consists in using progressive planning that adapts the detail of local plans according to local deadlines, and progressive negotiation that organizes conflicts between local plans into categories and solves them progressively from a mandatory category to an optional one. This structuration of conflicts in categories contributes in solving first, the most important conflicts in order to guarantee, when it is possible, the coordination of the mandatory part of the plan before the deadline. Our negotiation model is based on the modified PGP (Partial Global Planning) approach, named Partial Global Progressive Planning (PGPP), which is an incremental strategy to insert partial local plans progressively one by one. This strategy consists in discarding optional partial local plans of an agent when a deadline is exceeded or global consistency is violated. We show that this approach reduces the costs of detecting and solving conflicts. This approach can be seen as a step towards the application of contract-net type systems to real-world problems.

In this paper we present an approximation method based on discrete Hopfield neural network (DHNN) for solving temporal constraint satisfaction problems. This method is of interest for problems involving numeric and symbolic temporal constraints and where a solution satisfying the constraints of the problem needs to be found within a given deadline. More precisely the method has the ability to provide a solution with a quality proportional to the allocated process time. The quality of the solution corresponds here to the number of satisfied constraints. This property is very important for real world applications including reactive scheduling and planning and also for over constrained problems where a complete solution cannot be found. Experimental study, in terms of time cost and quality of the solution provided, of the DHNN based method we propose provides promising results comparing to the other exact methods based on branch and bound and approximation methods based on stochastic local search.

This paper presents a language for image processing-analysis called ElKONES. The language grammar is derived from the C language with extensions for efficient and flexible image processing. The EIKONES language provides to the user flexibility and friendliness for image processing which are not available in other image processing tools or conventional languages or function libraries. The basic idea behind EIKONES, is the consideration of the image processing algorithms as objects and the appropriate development of a formal grammar for its actual implementation. EIKONES provides a high level language for image processing with features such as learning capability, statistical and region analysis, object tracking. The way in which image operations are specified facilitates a natural language interface for use in a voice-activated system because multiple image operations are mapped to a single image word.

Temporal Web usage mining involves application of data mining techniques on temporal Web usage data to discover temporal usage patterns, which describe the temporal behavior of users on the Internet Web site, to understand the temporal users' behavior during different time slices. Clustering and classification are two important functions in Web mining. Classes, and associations in Web mining do not necessarily have crisp boundaries. Therefore the conventional clustering techniques became unsuitable to find such clusters and associations, where these conventional classification algorithms provide crisp classes, which are not suitable in real world applications. This gives the chance of using the non-conventional clustering techniques as fuzzy and rough sets in Web mining clustering applications. Recent research introduced the adaptation of Kohonen SOM based on the properties of rough sets theory to find the interval set clusters for the users on the Internet. This paper introduces the comparison between the latest snapshot Web usage mining and the temporal Web usage mining, and. the comparison between the temporal Web usage mining using the conventional Kohonen SOM and the modified Kohonen SOM based on the properties of sets theory.

Combination of logic-based artificial intelligence with virtual reality in intelligent agent systems is an approach not extensively sought after to date. It is our belief that significant gain is to be expected if the technical challenges involved are overcome. In this paper, we describe the mVlTAL intelligent agent system, which is our latest effort towards this direction. The system is a contemporary intelligent agent system with applications in numerous areas, including intelligent virtual environments and formal artificial intelligence research. The system focuses largely on logic-based approaches, which are present in almost every aspect of it, including modeling, knowledge representation, definition of agent behaviors and inter-agent communication. In addition, virtual manifestation of the world and agents is also an inherent characteristic of the system. The system, even if still in a development and evaluation stage, has already been employed in experimental and educational applications, demonstrating the potential benefits of such an approach.

Contrary to much of the research in machine learning where there is a concentration on problems with relatively small volume of data, one of the main challenges of the today's data mining systems is their ability to handle data that is substantially larger than available main memory on a single processor. In this paper, we present a distributed technique that consists in combining the partial results of different classifiers supplied, in parallel, with different subsets of data. It is, actually, a two-phase process. At first, a number of classifiers are trained, each with a different subset of the data. Then, the trained classifiers are used in the construction of a new training data set which has exactly the same format as, but is substantially smaller than, the initial one. The new data set is used to train the final classifier through an iterative process, which is guided by a threshold concerning the size of the data set and the achieved increase in accuracy. We present extensive empirical tests, which demonstrate that the proposed technique significantly reduces time complexity, usually at the expense of a lower accuracy, compared to a single classifier supplied with all the data.

In this article a Radial Basis Function Network (RBFN) approach for fast and efficient computation of inverse continuous time variant functions is presented. The approach is based on using a novel RBFN approach for computing inverse continuous time variant functions via a damped least squares formulation and also on a non-conventional implementation of an original approach for singularities prevention and conditioning improvement. The singularities avoidance approach in turn consists on establishing some characterizing matrices, in order to obtain a performance index and a null space vector, and then properly including it in the overall RBFN approach.

In this work, we develop a neural model to solve causal reasoning problems (said also abduction) in the open, independent and incompatibility classes. We model the reasoning process by a single and global energy function using cooperative and competitive neural computation. The update rules of the distinct connections of the network are derived from its energy function using gradient descent techniques. Simulation results reveal a good performance of the model.

The huge and complicated plants such as nuclear power stations are likely to cause the operators to make mistakes due to a variety of inexplicable reasons and symptoms in case of emergency. That's why the prevention system assisting the operators is being developed for. First of all, I suggest an improved fuzzy diagnosis. Secondly, I want to demonstrate that a classification system of nuclear plant's accident investigating the causes of accidents foresees possible problems, and maintains the reliability of the diagnostic reports in spite of improper working in part.

In the event of emergency in a nuclear plant, a lot of operational steps enable the operators to find out what caused the problems based on an emergent operating plan. Our system is able to classify their types within twenty to thirty seconds. As so, we expect the system to put down the accidents right after the rapid detection of the damage control-method concerned.

In this paper we describe an approach to define flexible Knowledge Extraction Systems able to deal with the inherent vagueness and uncertainty of the Extraction process. We also present a short survey of fuzzy and semantic approaches to knowledge extraction. The goal of such approaches is to address if and how some approaches met their goal.

This paper describes a learning/adaptive approach to automatically building knowledge bases for information extraction from text based web pages. A frame based representation is introduced to represent domain knowledge as knowledge unit frames. A frame learning algorithm is developed to automatically learn knowledge unit frames from training examples. Some training examples can be obtained by automatically parsing a number of tabular web pages in the same domain, which greatly reduced the amount of time consuming manual work. This approach was investigated on ten web sites of real estate advertisements and car advertisements and nearly all the information was successfully extracted with very few false alarms. These results suggest that both the knowledge unit frame representation and the frame learning algorithm work well, domain specific knowledge bases can be learned from training examples, and the domain specific knowledge base can be used for information extraction from flexible text-based semi-structured Web pages on multiple Web sites. The investigation of the knowledge representation on five other domains suggests that this approach can be easily applied to other domains by simply changing the training examples.

Several advances have been made on managing multimedia databases as well as on data mining. Recently there is active research on mining multimedia databases. This paper provides an overview of managing multimedia databases and then describes issues on mining multimedia databases. In particular mining text, image, audio and video data are discussed.