a b s t r a c tNeural network has been applied in several classification problems such as in medical diagnosis, handwriting recognition, and product inspection, with a good classification performance. The performance of a neural network is characterized by the neural network's structure, transfer function, and learning algorithm. However, a neural network classifier tends to be weak if it uses an inappropriate structure. The neural network's structure depends on the complexity of the relationship between the input and the output. There are no exact rules that can be used to determine the neural network's structure. Therefore, studies in improving neural network classification performance without changing the neural network's structure is a challenging issue. This paper proposes a method to improve neural network classification performance by constructing a linear model based on the Kalman filter as a post processing. The linear model transforms the predicted output of the neural network to a value close to the desired output by using the linear combination of the object features and the predicted output. This simple transformation will reduce the error of neural network and improve classification performance. The Kalman filter iteration is used to estimate the parameters of the linear model. Five datasets from various domains with various characteristics, such as attribute types, the number of attributes, the number of samples, and the number of classes, were used for empirical validation. The validation results show that the linear model based on the Kalman filter can improve the performance of the original neural network.

Abstract-Image segmentation plays an important role in automatic visual inspection of food product using computer vision system. However, segmentation of food product image is not easily performed if the image has low contrast with its background or the background in acquired image is not homogeneous. This paper proposes a method for automatic food product image segmentation using Sobel operator and k-means clustering. Sobel operator was used to determine region of interest (ROI). k-means clustering was then used to separate object and background in ROI. The area outside ROI was considered as background. The proposed method has been validated using 100 images of food product from ten different types. The experimental results show that the proposed method achieves good segmentation result.

Augmented reality (AR) technology is a technique that combines the real world and the virtual world digitally using mobile devices. Mobile AR technology is expected to help Malaysian tourists who have difficulties to understand the Thai language when visiting the country. Hence, a prototype called ARThaiMalay translator was developed to translate printed Thai food menu to Malay language. The objective of this study is to design a food menu translation framework from Thai to Malay language based on mobile AR, develop a translator application and to test the effectiveness of the translator application. The prototype consists of three main components which are translation based on optical character recognition (OCR) technology, dictionary development using SQLite database and display data from the local database. Evaluation of the developed application shows its effectiveness to perform translation of Thai text with certain features to Malay language.

Volume measurement plays an important role in the production and processing of food products. Various methods have been proposed to measure the volume of food products with irregular shapes based on 3D reconstruction. However, 3D reconstruction comes with a high-priced computational cost. Furthermore, some of the volume measurement methods based on 3D reconstruction have a low accuracy. Another method for measuring volume of objects uses Monte Carlo method. Monte Carlo method performs volume measurements using random points. Monte Carlo method only requires information regarding whether random points fall inside or outside an object and does not require a 3D reconstruction. This paper proposes volume measurement using a computer vision system for irregularly shaped food products without 3D reconstruction based on Monte Carlo method with heuristic adjustment. Five images of food product were captured using five cameras and processed to produce binary images. Monte Carlo integration with heuristic adjustment was performed to measure the volume based on the information extracted from binary images. The experimental results show that the proposed method provided high accuracy and precision compared to the water displacement method. In addition, the proposed method is more accurate and faster than the space carving method.

A statistical approach to automatic speech recognition using the atomic speech units constructed from overlapping articulatory features Abstract. A novel method is proposed to recognize the Arab/Jawi and Roman digits. This new method is based on features from the triangle geometry, normalized into nine features. The features are used for zoning which results in five and 25 zones. The algorithm is validated by using three standard datasets which are publicly available and used by researchers in this field. The first dataset is HODA that contains 60,000 images for training and 20,000 images for testing. The second dataset is IFHCDB. This dataset has 52,380 isolated characters and 17,740 digits. Only the 17,740 images of digits are used for this research. For the roman digit, MNIST are chosen. MNIST dataset has 60,000 images for training and 10,000 images for testing. Supervised (SML) and Unsupervised Machine Learning (UML) are used to test the nine features. The SML used are Neural Network (NN) and Support Vector Machine (SVM). Whereas the UML uses Euclidean Distance Method with data mining algorithms; namely Mean Average Precision (eMAP) and Frequency Based (eFB). Results for SML testing for HODA dataset are 98.07% accuracy for SVM, and 96.73% for NN. For IFHCDB and MNIST the accuracy are 91.75% and 93.095% respectively. For the UML tests, HODA dataset is 93.91%, IFHCDB 85.94% and MNIST 86.61%. The train and test images are selected using both random and the original dataset's distribution. The results show that the accuracy of proposed algorithm is over 90% for each SML trained datasets where the highest result is the one that uses 25 zones features.

No abstract

A chain of three oscillators A, B, C with nearest neighbour coupling, is considered. It is shown that the correlations between A, C (which are not coupled directly) can be stronger than the correlations between A, B. Also in some cases various witnesses of entanglement show that A, C are entangled but they cannot lead to any conclusion about A, B.

Abstract. Ontology learning (OL) is the computational task of generating a knowledge base in the form of an ontology, given an unstructured corpus in natural language (NL). While most works in the field of ontology learning have been primarily based on a statistical approach to extract lightweight OL, very few attempts have been made to extract axiomatic OL (called heavyweight OL) from NL text documents. Axiomatic OL supports more precise formal logic-based reasoning when compared to lightweight OL. Lexico-syntactic pattern matching and statisticsal one cannot lead to very accurate learning, mostly because of several linguistic nuances in the NL. Axiomatic OL is an alternative methodology that has not been explored much, where a deep linguistics analysis in computational linguistics is used to generate formal axioms and definitions instead of simply inducing a taxonomy. The ontology that is created not only stores the information about the application domain in explicit knowledge, but also can deduce the implicit knowledge from this ontology. This research will explore the English translation of the meaning of Quranic texts.