BUJICT 2014

1-6

Authors: Hla Myo Tun

Creating computer generated caricature from given photograph will be a great source of entertainment to all people because caricatures are funny, humorous and satirical. To generate caricature it requires artificial intelligence to capture the non-linear relationships between photograph and drawn caricature. It means that computer has to think itself what will be caricature for the given photograph. A formalization of this idea, Exaggerating the Difference from the Mean (EDFM) is widely accepted among caricaturists to be the driving factor behind caricature generation. As artificial intelligence is based on neural network, the key research is required to investigate neural network and design its implementation. No attempt has been taken in the past to identify these distinct drawing styles. Yet the proper identification of the drawing style of an artist will allow the accurate modeling of a personalized exaggeration process, leading to fully automatic caricature generation with increased accuracy. In this paper the author provides experimental results and detailed analysis to prove that a Cascade Correlation Neural Network (CCNN) can be successfully used for transforming caricature generation from photos.

10-13

Authors: Saifuddin Ahmed Jilani and M.Ayub Khan Yousuf Zai

In this communication, we use observations of solar flare duration for calculation of intensity of solar particles emitted at Pakistan atmospheric region over a period of 1979-2006, to predict local behavior with proper degree of accuracy. Actually short records not fully describe the diurnal cycle but different spectral technique like Periodogram and spectral density are helpful in disclosing the true picture behind any physical phenomena. We find hidden periodicities and pin-point the peak frequencies in the time series during active sun by applying confidence limits to the peaks of these spectral estimates. The Fourier coefficients and spectral energy for different portion of the time series have also been calculated.

18-25

Authors: Pang Ying Han, Andrew Teoh Beng Jin, Fazly Salleh Abas and Ooi Shih Yin

Locally Linear Embedding (LLE) is an unsupervised non-linear manifold learning method, which has spurred increased interest in face recognition research recently. However, it is commonly known that a supervised method that considering the class-specific information always outperforms the unsupervised one, especially in biometric recognition task. In this paper, we propose a supervised LLE technique, known as class-label Locally Linear Embedding (cLLE). cLLE aims to discover the non-linearity of high-dimensional data by minimizing the global reconstruction error of the set of all local neighbors in the data set. cLLE method is using user class-specific information in neighborhoods selection and thus preserves the local neighborhoods. Since the locality preservation is correlated to the class discrimination, the proposed cLLE is expected superior to LLE in face recognition. Experimental results on three face databases demonstrate the success of the proposed technique.

37-42

Authors: Shahzad Ahmed

There have been a number of methods presented by various researchers for traffic prediction, some of which involve modeling the problem of traffic prediction as a time series. It has been observed that Artificial Neural Networks (ANN) perform better than statistical methods for time series forecasting. The network performance and complexity varies with the choice of algorithm used. Back propagation (BPNN) has been used to predict IP traffic with a fair degree of accuracy but as the prediction interval increases and the inputs change drastically the forecasting accuracy suffers [9]. This paper discusses the use of Focused Time Delay Feed Forward Neural Network architecture to predict IP traffic patterns and overcome the short comings of back propagation neural networks when used for traffic forecasting along with improvements to the BPNN by using additional inputs like holidays and maintenance downtimes.

53-55

Authors: S.M. Zia-ul-Haque and M.Ayub Khan Yousuf Zai

In this communication, the effect. of ozone layer depletion on the chlorophyll of the plants under observation has been studied. Ozone absorbs the specific range of ultraviolet radiation (UV) reaching the earth’s surface. Data for two samples was selected for analysis. One set of data used to measure chlorophyll fluorescence before expose to UV termed as number of chlorophyll fluorescence before exposure (NOCFBEX) and second set of data utilized to measure chlorophyll fluorescence solar exposure that is indicated by number of chlorophyll fluorescence solar exposure (NOCFSEX). It has been observed that the process before exposure is quite acceptable, that the chlorophyll is maximum and varying with the number of time of observations. In the second case, the minimum chlorophyll has been exhibited due to the disorder in the chlorophyll nature and the process becomes non-stationary. Quantification of the chlorophyll fluorescence of the leaves of the plant under observations has been interpreted. As a result of UV-B exposure trend equations exhibit the existence of non-stationary and stationary in the leaves.