International Journal on Artificial Intelligence Tools

There are several reasons to expect that recognising word order errors in a text will be a difficult problem, and recognition rates reported in the literature are in fact low. Although grammatical rules constructed by computational linguists improve the performance of a grammar checker in word order diagnosis, the repairing task is still very difficult. This paper describes a method to repair any sentence with wrong word order using a statistical language model (LM). A good indicator of whether a person really knows a language is the ability to use the appropriate words in a sentence in correct word order. The "scrambled" words in a sentence produce a meaningless sentence. Most languages have a fairly fixed word order. This paper introduces a method, which is language independent, for repairing word order errors in sentences using the probabilities of most typical trigrams and bigrams extracted from a large text corpus such as the British National Corpus (BNC).

Automated offline handwritten character recognition involves the development of computational methods that can generate descriptions of the handwritten objects from scanned digital images. This is a challenging computational task, due to the vast impreciseness associated with the handwritten patterns of different individuals. Therefore, to be successful, any solution should employ techniques that can effectively handle this imprecise knowledge. Fuzzy Logic, with its ability to deal with the impreciseness arisen due to lack of knowledge, could be successfully used to develop automated systems for handwritten character recognition. This paper presents an approach towards the development of a customizable fuzzy system for offline handwritten character recognition.

Building agents that negotiate on behalf of their owners aiming to maximise their utility is a quite challenging research field in the artificial intelligence domain. In this paper, such agents are enhanced with techniques based on neural networks (NNs) to predict their opponents' negotiation behaviour, thus achieving more profitable results and better resource utilization. The NNs are used to early detect the cases where agreements are not achievable, supporting the decision of the agents to withdraw or not from the negotiation threads. The designed NN-assisted negotiation strategies have been evaluated via extensive experiments and are proven to be very useful.

The behavioral-biometrics methods of writer identification and verification have been considered as a research topic for many years. However, many writer identification and verification methods have been designed based on English handwriting properties, but because of many differences between English and Persian handwriting and the challenges facing Persian handwriting analysis, designing such methods has many interests in Persian yet. In this paper, we have presented a fully text-independent and texture based method for identifying writers of Persian handwritten documents. As a result of special properties of Persian handwriting, a modified version of Gabor filter that is called Extended Gabor (XGabor) filter has been used to extract the features. An MLP (Multi Layer Perceptron (Node)) neural network and a K-NN classifier have been employed to classify the extracted features. In the evaluation phase, an exhaustive database of Persian handwritten documents was prepared and the method applied on. The experimental results showed that the accuracy of proposed method is about 97% and it is competitive with others. We believe that the proposed method may be extended to identify writers in other languages by adjusting some parameters.

Image analysis represents an invaluable tool in processing and analyzing biological data in an expeditious and reliable way. This paper describes the design and implementation of an image analysis system for the automatic assessment of the growth of Heliothis-zea insects from color images. Specifically, the Heliothis zea is a corn earworm eating corn crops. Biotech researchers are interested in developing insecticidal bio-toxins with the best performance to kill or stunt the growth of the pest. Currently, assessing the effectiveness of different bio-toxin solutions is done mostly manually by biotech experts. The goal of this study is to investigate the use of image analysis for automating and improving the efficiency of this process. In this context, we have developed a prototype system for assessing insect growth from color images which contains three main stages: (1) insect segmentation from background, (2) region processing and feature extraction, and (3) categorization of insect growth. A probabilistic model based on mixtures of Gaussians has been adopted to segment the insect images. Also, back-propagation neural networks have been trained to classify the instar stage and life stage. The proposed system has been evaluated on a set of real images.

This paper addresses the problem of learning to classify texts by exploiting information derived from clustering both training and testing sets. The incorporation of knowledge resulting from clustering into the feature space representation of the texts is expected to boost the performance of a classifier. We present an empirical study of the proposed algorithm on a variety of datasets. We attempt to answer various questions such as the effect of independence or relevance amongst features, the effect of the size of the training set and the effect of noise. The results are encouraging, revealing that information resulting from clustering can create text classifiers of high-accuracy.

Based on clonal selection principle, this paper proposes an immune-inspired evolution strategy (IIES) for constrained optimization problems with two improvements. Firstly, in order to enhance global search capability, more clones are produced by individuals that have far-off nearest neighbors in the less-crowed regions. On the other hand, immune update mechanism is proposed to replace the worst individuals in clone population with the best individuals stored in immune memory in every generation. Therefore, search direction can always focus on the fittest individuals. These proposals are able to avoid being trapped in local optimal regions and remarkably enhance global search capability. In order to examine the optimization performance of IIES, 13 well-known benchmark test functions are used. When comparing with various state-of-the-arts and recently proposed competent algorithms, simulation results show that IIES performs better or comparably in most cases.

Support Vector Machines (SVMs) are based on the concept of decision planes that define decision boundaries, and Least Squares Support Vector (LS-SVM) Machine is the reformulation of the principles of SVM. In this study a diagnosis on a BUPA liver disorders dataset, is conducted LS-SVM with the Taguchi method. The BUPA Liver Disorders dataset includes 345 samples with 6 features and 2 class labels. The system approach has two stages. In the first stage, in order to effectively determine the parameters of the kernel function, the Taguchi method is used to obtain better parameter settings. In the second stage, diagnosis of the BUPA liver disorders dataset is conducted using the LS-SVM classifier; the classification accuracy is 95.07%; the AROC is 99.12%. Compared with the results of related research, our proposed system is both effective and reliable.

In multiple attribute decision making (MADM) problems, it is usual that no single alternative works best for all performance attributes, so it is difficult to select the best from among multiple available alternatives, especially in the situation that the attribute weights are completely unknown. This research propose a game theory-based approach (GMADM) which incorporates static strategic game theory into MADM problems to derive the attribute weights, and then utilize the weight arithmetic average (WAA) operator to aggregate the attribute values corresponding to each alternative and rank alternatives by means of aggregated information. In GMADM, each attribute is regarded as a player taking part in the game, and the player's strategy is to select a value from interval [0,1] to assign corresponding attribute weight, and the player's utility is defined as the agreement between the ranking of alternatives determined by the aggregated information and the one determined by the attribute values. When the game is in equilibrium status, the strategy profile is the best attribute weights which make each player have good utilities. Moreover, the equilibrium solution of game and the resolution method for the MADM problem without weight information have also been developed. Finally, the result of proposed approach for a practical MADM problem and its comparisons with one of other methods are given.