Ramkhamhaeng Research Journal of Sciences and Technology

Incremental Learning Probabilistic State Machine for Symbolic Sequences Classification

jittakorn pullpothong — 2021-06-29

Symbolic sequence classification can be used in a variety of applications such as DNA sequences analysis, intrusion detection, electrocardiography (ECG) analysis. The convention methods can be applied to this field for example, probabilistic language model, support vector machine, and artificial neural network. However, learning unknown words in very long sequences such as DNA sequences need fix size of imitation words. Consequently, the probability and position of original words are distorted which can lead to incorrect results in long term. Moreover, learning update process may cause confliction with the former fix size knowledge. In this work, to optimise the probability and the position of original words as possible while the direction of data is still tractable, we propose a novel probabilistic language model using Markov assumption with variable memory length considered by unique and repeated substring. We propose an on-line algorithm to build the model with time and space complexity . The experiment of classification applied to DNA sequences of promoter and non-promoter bacteria E. Coli. The error of classification is only 3.77 % which is satisfied compared to the other methods.

Study on the Conditions for Emerging of Carcinogenic Benzene in Pasteurized 25 % Orange Juice Containing Benzoic and Ascorbic Acid

chulaporn lertborwornwong — 2021-06-29

The objective of this experiment was to study the possibility of benzene formation at different

conditions used in pasteurized 25% orange juice production and storage : benzoate or ascorbic acid alone,

benzoate and ascorbic acid together, pH, heating temperature and time, storage conditions(temperature, time

and light exposure),metal ions and chelating agents were studied to determine how high of benzene could

be formed if starting from one optimum condition for benzene formation until all conditions are optimized

then we will know what factors play a major role to reduce benzene level within safety limit . An addition

of 200 ppm sodium benzoate and 120 ppm ascorbic acid to 25% orange juice, pH 3.7, pasteurized at 100

^OC 10 min, stored at 30 ^OC 24 hrs., did not induce benzene formation. The amount of benzene increased

from 0 at pH 3.7 (commercial practice) to 5.6 ppb at pH 2. Increased processing temperature from 90 ^OC to

100 ^OC for 20 minutes, found benzene increased from 0 to 8.82 ppb and processing time from 10 to 20

minutes at 100 ^OC, found benzene increased from 5.32 to 8.82 ppb. Higher storage temperature caused

more benzene formation than longer storage time. Light exposure did not increase benzene formation

. Benzene formation in product added only sodium benzoate or ascorbic acid was less than sodium benzoate

together with ascorbic acid (stored at 45 ^OC). The addition of chelating agent (disodium EDTA and sodium

hexametaphosphate, SHMP) can decrease benzene formation.

Investigation for the Antibacterial Activity of Five Citrus Fruits’ Peel Extracts

Wimon Chanchaem — 2021-06-29

Bacterial infections are very important problems of world population. Many countries have to pay large budgets dealing with these diseases. However, there were lots of studies currently reported that fruits and folk plants were rich sources of medicinal compounds those have high potential as antimicrobial agents for alternative treatments. In this study, fruit peels of five citrus fruits including Citrus maxima, Citrus reticulata Blanco, Citrus reticulata Blanco cv. Sainampueng, Citrus hystrix DC and Citrus aurantifolia were extracted by using four methods including boiling, maceration in absolute ethanol, blending of fruit peels powder in water and blending of fresh fruit peels in water. The antibacterial activities of citrus fruits’ peels were evaluated using the agar disk diffusion method towards Escherichia coli, Staphylococcus aureus and Pseudomonas aeruginosa. The results showed that crude extracts of these five fruits’peels from ethanolic extraction had antibacterial activity to S. aureus and those using dry powder blending extraction had antibacterial activity to P. aeruginosa. No inhibition zone was found in all five extracts that used boiling extraction and fresh peel blending extraction.This study also showed that the activities of those extracts depended on many factors such as species of fruits, solvent, and the method of extraction. This preliminary study demonstrated that these citrus fruits’ peels contained some interesting levels of antibacterial components. Further study for the type and quantity of those important substances should be performed. This will lead to proper applications in pharmaceuticals or other antimicrobial products in future.