UPLB ICS Peak One

Finite State Automata and Machines Mobile (FSAM): A Visualization Centric Finite State Machine Simulator accepting Diagrammatic Input for Mobile Devices
Von Vincent O. Vista, Jaime M. Samaniego

Formal languages and automata theory are fundamental pillars of computer science. Educators recognised early on that automata theory is difficult to teach due to its theoretical nature, prompting the development and application of pedagogical tools for teaching and learning. There are numerous PC and web-based simulators, but the mobile platform is lacking, and all of the available simulators have either: only basic features; being Descriptive Language Based or Visualization Centric and accepting Structured Input in paradigm; or having a poor UI and UX when evaluated in accordance with UI guidelines and UX principles. This resulted in the development of Finite State Automata and Machines mobile (FSAM), which offers diagramming and simulation capabilities for finite state automata, pushdown automata, and turing machines. Moreover, it offers step and bulk simulations, file I/O with cross-compatibility with JFLAP, NFA conversion, and DFA minimization. There are no public C# libraries for automata building and simulation at the moment, and FSAM was built from the ground up. User evaluation was conducted using the Usefulness, Satisfaction, and Ease of use (USE) questionnaire with respondents who have used automata simulators before. Using a 7-point likert scale and calculating the mean scores, the results show that Usefulness received 6.40, Ease of Use received 6.36, Ease of Learning received 6.48, and Satisfaction received 6.53. All of the scales received a mean score of 6 (agree) to 7 (strongly agree), indicating that respondents overall agree-strongly agree with FSAM’s usefulness, ease of use, ease of learning, and satisfaction. Furthermore, the system is built and designed to be modular, allowing FSAM to be expanded and extended to other finite state machine applications and functions.

Published on July 2023, Search Score: 0, [BibTeX]
SApp: Data Visualization and Sentiment Analysis Tool From Twitter Data
Lorenzo Miguel B. Villapando, Jaime M. Samaniego

SApp, short for Sentiment Analysis Application, provides tools for data gathering, visualization, and sentiment analysis for Twitter. Sentiment Analysis and Twitter data visualization are tools that may be used to guide business people, as well as business-minded people, in creating plans and monitoring public sentiment on products or topics of interest. For SApp, the web application detailed in the paper, Twitter data was extracted using SNScrape, a powerful scraper for social networking services. Visualization utilized Plotly and Mathplotlib, presented using Streamlit. Sentiment analysis used ”twitter-roberta-basesentiment-latest,” a RoBERTa-base model trained on around 124 million tweets from 2018-2021. In addition, the libraries Deep Translator and Detect Language were used. The web application was tested using the System Usability Scale on a total of 16 respondents. This resulted in a score of 80, meaning that the system is usable and performs the task as intended, even for non-technical users.

Published on July 2023, Search Score: 0, [BibTeX]
OptiCare: A Mobile Vision-Screening Application for Teleoptometry
Ellison Maurice C. Paguagan, Monina Gazelle Charina B. Carandang

OptiCare is a mobile vision-screening application designed to provide teleoptometry services to individuals who have limited access to eye care services. This study presents the development, implementation, and evaluation of OptiCare as a solution to the growing need for accessible and affordable eye care, including the use of near-visual acuity testing, central vision testing, and color vision testing. The usability of the application was evaluated by using the System Usability Scale (SUS) questionnaire, which yielded an average SUS score of 87.25 obtained from patients, indicating a favorable level of usability, and an average SUS score of 92.5 obtained from optometrists, reflecting an even higher level of usability.

Published on July 2023, Search Score: 0, [BibTeX]
HealthyPeeps: A Philippine Online Platform for Nutrition and Healthy Eating Enthusiasts
Crisneil Mae O. Musa, Toni-Jan Keith P. Monserrat

As the number of overweight and obese individuals in the Philippines grows, the risk of premature death and health complications also increases. With this in mind, this study developed HealthyPeeps, an online platform designed to provide Filipinos, particularly those interested in healthier lifestyles, with a central place to access and share health and nutrition information. A survey was given to 15 respondents consisting of UPLB students, professors, and alumni to determine the usability of the application. Based on the results, HealthyPeeps achieved a System Usability Scale score of 82, indicating that the application is user-friendly and efficient. HealthyPeeps serves as more than an information hub - it’s a catalyst for positive health changes, encouraging better dietary habits and nutrition awareness among users. Offering culturally relevant content, it empowers users to make informed food choices, potentially improving the overall health of the Filipino community. In summary, HealthyPeeps holds promise as an impactful tool for promoting healthier lifestyles among Filipinos, blending cultural relevance with health promotion.

Published on July 2023, Search Score: 0, [BibTeX]
Improving Tomato Pest Image Classification with Transformative and Generative Data Augmentation
Jose Enrique R. Lopez, Concepcion L. Khan

Datasets in crop image classification often suffer from limitations in quantity, balance, and fidelity to real-life conditions. Models trained using these datasets overfit and offer poor generalizability to actual samples of infected leaves. We evaluate the effectiveness of transformative and generative augmentation in enhancing a limited pest dataset to improve model accuracy. One control and two experimental setups were used. The control setup provided no augmentation to the limited pest dataset. The first experimental setup used random transforms in brightness, rotation, and color shift to expand the dataset. The second experimental setup utilized a generative adversarial network to generate fake images of pests to augment the limtied training set. Three separate models were trained for each setup, getting their accuracy scores as a measure of model skill. To gain a reliable score, the experiment was repeated 1000 times. At α = 0.05, the differences in the mean classification accuracy scores of the three experimental setups were significantly different. In particular, transformative augmentation performed the best in boosting classification accuracy (α = 0.033) as it helped reduce noise and biases in background, lighting, and angle.

Published on July 2023, Search Score: 0, [BibTeX]
Next page >



Results found in 0.0015022754669189453 seconds..