TEMATIK

Penerapan Metode Stable Diffusion Dengan Fine Tuning Untuk Pola Endek Bali

Ni Luh Wiwik Sri Rahayu Ginantra — 2024-11-05

Endek Bali fabric is a cultural heritage of Bali renowned for its traditional decorative motifs, including floral, fauna, patra, and diamond patterns. Although rich in cultural value, artisans often face challenges in creating new designs that align with market trends while preserving cultural authenticity. Artificial Intelligence (AI) technology, particularly text-to-image generation models, offers a solution to this issue by streamlining the design process and enabling the exploration of new motifs. The Stable Diffusion model, introduced by Stability.AI in 2022 and open source, can be utilized to generate Endek Bali patterns through Fine Tuning techniques. Fine Tuning allows the model to be adapted to specific domains, enhancing its performance in generating textile patterns based on textual descriptions. This study aims to apply the Stable Diffusion model and Fine Tuning techniques to create new patterns and motifs. By using this model, it is hoped that innovative designs can be produced while maintaining the authenticity and local cultural values of Bali. The research demonstrates that the Fine-Tuned Stable Diffusion model is effective in creating Endek Bali patterns with high accuracy, as evaluated by Clip Similarity, with the highest scores achieved for Floral Patterns (92.43), followed by Decorative (free-form motifs) Floral (88.77), Decorative (free-form motifs) Geometric (87.94), and Decorative (free-form motifs) (85.79). These findings indicate the model’s flexibility and effectiveness in producing intricate textile designs, enabling designers and artisans to generate complex and innovative patterns solely from textual descriptions while preserving Bali’s cultural values.

Exploration of Acceptance Factors of Online Learning Platform: A Theory of Planned Behavior Perspective

Adi Dharma Putra — 2024-12-09

E-learning is recognized as a form of education facilitated through the application of information and communication technologies. The successful implementation of e-learning is influenced by numerous factors. This research aims to identify the determinants of online learning platform acceptance through the lens of the Theory of Planned Behavior (TPB). The study focuses on three independent variables: attitudes, subjective norms, and perceived behavioral control. Positioned as explanatory research with a quantitative emphasis, data were gathered from active students who met specific criteria and were subsequently given access to an online questionnaire. The research sample was obtained through a purposive sampling method. This study employs multiple linear regression analysis, executed through computational means. The hypothesis testing results indicate that two hypotheses—subjective norms on intention and intention on behavior—have a positive and significant influence, suggesting that perceptions from one's social environment can significantly shape users' intentions to engage with online learning platforms. However, the remaining three hypotheses reveal a positive yet insignificant effect, indicating that the variables of behavior and attitude control do not significantly impact the use of online learning platforms. Collectively, the four variables account for only 43.7% of the behavioral outcomes, with the remaining 56.3% likely influenced by other factors beyond the scope of this research.

E-learning dikenal sebagai bentuk pendidikan yang difasilitasi melalui penerapan teknologi informasi dan komunikasi. Keberhasilan implementasi e-learning dipengaruhi oleh banyak faktor. Penelitian ini bertujuan untuk mengidentifikasi faktor-faktor penentu penerimaan platform pembelajaran online melalui lensa Theory of Planned Behavior (TPB). Penelitian ini berfokus pada tiga variabel independen: sikap, norma subjektif, dan kontrol perilaku yang dirasakan. Diposisikan sebagai penelitian eksplanatori dengan penekanan kuantitatif, data dikumpulkan dari mahasiswa aktif yang memenuhi kriteria tertentu dan kemudian diberikan akses ke kuesioner online. Sampel penelitian diperoleh melalui metode purposive sampling. Penelitian ini menggunakan analisis regresi linier berganda, yang dilakukan dengan cara komputasi. Hasil pengujian hipotesis menunjukkan bahwa dua hipotesis - norma subjektif terhadap niat dan niat terhadap perilaku - memiliki pengaruh yang positif dan signifikan, yang menunjukkan bahwa persepsi dari lingkungan sosial seseorang dapat secara signifikan membentuk niat pengguna untuk terlibat dengan platform pembelajaran online. Namun, tiga hipotesis lainnya menunjukkan pengaruh yang positif namun tidak signifikan, mengindikasikan bahwa variabel perilaku dan kontrol sikap tidak secara signifikan mempengaruhi penggunaan platform pembelajaran online. Secara kolektif, keempat variabel tersebut hanya menjelaskan 43,7% dari hasil perilaku, dengan 56,3% sisanya kemungkinan dipengaruhi oleh faktor-faktor lain di luar cakupan penelitian ini.

Kombinasi Metode SVM Dengan Optimasi SMOTE Terhadap Ulasan Pengguna Layanan Streaming

Ni Putu Anik Juniantini — 2024-12-09

The development of digital technology has changed media consumption patterns in Indonesia, with streaming services like Disney+ Hotstar becoming increasingly popular. Since its launch in 2020, Disney+ Hotstar has offered exclusive content from Marvel, Pixar, Disney, National Geographic and others, quickly capturing the market in Indonesia. However, this service is not without controversy, particularly regarding certain content deemed to conflict with social values in Indonesia. Additionally, service quality, subscription prices, and content availability are also concerns for users. The difference in ratings on the Play Store (1.7) and the App Store (4.8) indicates a disparity in user satisfaction between the two platforms. This study aims to analyze user sentiment towards Disney+ Hotstar, particularly regarding reviews on the Play Store and App Store. Using a classification model with Support Vector Machine (SVM), optimized with the Synthetic Minority Over-Sampling Technique (SMOTE) to address data imbalance. Based on the analysis of 1,650 datasets, user sentiment tends to be neutral, as measured using the Vader Lexicon. The method testing results show that SMOTE optimization can improve the performance of the SVM model, with an accuracy increase of +0.7 on Play Store reviews from 0.67 to 0.74, and an accuracy increase of +0.11 on App Store reviews from 0.72 to 0.83 In conclusion, the SVM method optimized with SMOTE has proven effective in improving the accuracy of sentiment classification in user reviews of Disney+ Hotstar.

Incorporation of Real Experiences and Artifacts As an Online Learning Intervention

Erda Guslinar Perdana — 2024-12-16

Online learning has become a significant trend in recent years, but the lack of direct interaction between students and lecturers can reduce its effectiveness. This study addresses this challenge by incorporating real-world experiences—specifically two case studies, one real-world project, and simulations—alongside concrete artifacts such as software documentation, research reports, and test results. This research aims to improve the effectiveness of online learning in software verification and validation courses by incorporating real-world experiences and artifacts. A case study approach was used to test the effectiveness of this intervention by involving students from related courses. Data were collected through interviews, observations, and surveys, and analyzed qualitatively and quantitatively to understand the impact of the intervention on improving the learning experience and understanding of software verification and validation concepts. Qualitative analysis highlighted that the integration of these elements improved students' conceptual understanding and practical application skills. Quantitative results indicated a 30% increase in student engagement and a 25% improvement in understanding course material compared to prior cohorts. The findings underscore the importance of integrating real-world experiences and artifacts into online learning to promote deeper engagement, better comprehension, and more authentic learning experiences.

Analisis Pergerakan Lengan Tari Bedoyo Majapahit Berbasis Motion Capture

Anang Kukuh Adisusilo — 2024-12-16

The Bedoyo Majapahit dance is an Indonesian cultural heritage that combines elegance and meaning in every movement, particularly in its signature arm movements. This study focuses on exploring and analyzing arm movement patterns in this traditional dance using motion capture technology, aiming to document, understand, and scientifically reveal the beauty and dynamics of these movements. Positional data of the shoulders, elbows, and wrists from both sides were analyzed to calculate speed and elbow angles. Descriptive statistics, correlation analysis, and K-Means clustering were applied to identify dominant patterns. The results indicate a positive correlation between shoulder and elbow speeds on both sides, with correlation values of 0.71 and 0.84, highlighting symmetrical movements. Three movement clusters were identified: low, medium, and high, with the low cluster being the most dominant. The average speeds in the low cluster were 0.097 m/s for the shoulders, 0.092 m/s for the elbows, and 0.091 m/s for the wrists, reflecting the gentle characteristics of the dance. Meanwhile, the medium and high clusters exhibited higher speed values, particularly for the wrists in the high cluster at 1.888 m/s, indicating more dynamic movements. This study provides a quantitative understanding of the smooth and symmetrical movements in Bedoyo Majapahit dance, supporting cultural preservation through data-driven analytical approaches.

Manajemen Pertukaran Data Antar Kendaraan Menerapkan Protokol CoAP

Kadek Suar Wibawa — 2024-12-18

This study aims to design and develop a prototype model for inter-vehicle data exchange services using the LoRa Ra-02 Sx1278 radio module, ESP8266 microcontroller, and WiFi for internet connectivity. This system enables effective, inexpensive, and instant communication. In its implementation, we optimized simple data management to store the results of data exchanges on a database server connected to the internet using the Constraint Application Protocol (CoAP) communication protocol. The system is also capable of vehicle monitoring through a web-based application by integrating a GPS receiver. The application services and prototypes are designed to minimize costs and ease of implementation. Research and Development with a limited sample is the method applied in the development of this prototype. Testing results show that the prototype can exchange data at distances up to 500m with a success rate of 86%. The visualization of the stored vehicle location data can be monitored through the website application.

Model Prediksi Kepadatan Pariwisata Jawa Barat Menggunakan Metode Long Short-Term Memory with Temporal Attention

Nadya Safitri — 2024-12-20

This study aims to apply the Long Short-Term Memory Networks (LSTM) with Temporal Attention method in predicting tourism density in West Java tourist destinations. The problem faced is the uncertainty in estimating tourist density at various locations and times, which makes the management of tourism resources and facilities difficult. Therefore, this study is important to provide a tool that can help make more effective decisions in the tourism sector in West Java. The urgency of this study lies in the need for accurate and real-time tourist density predictions to support the management and development of tourist destinations in West Java. With the right prediction model, related parties can regulate capacity, optimize services, and avoid negative impacts such as excess capacity and crowds that have the potential to endanger visitors and the environment. The purpose of this study is to develop a tourism density prediction model that combines the distinctive features of LSTM with a temporal attention mechanism. This model aims to provide accurate and dynamic tourist density estimates, taking into account the temporal patterns of tourist visits in West Java. The model evaluation methods used in this study are RMSE and MAE, and the results of the model testing are that it has an RMSE value of 32208867.139 and an MAE value of 5099.219, and it is hoped that there will be a dataset with a long period after the covid mass where the dataset is free from abnormal events so that a more appropriate model is obtained.

Identifikasi Opini Publik Terhadap Kendaraan Listrik dari Data Komentar YouTube: Pemodelan Topik Menggunakan BERTopic

Kristine Angelina Simanjuntak — 2024-12-20

The Indonesian government is encouraging the transition to electric vehicles to reduce the use of fossil fuels and the negative environmental impact. This transition sparked controversy because Indonesia is still heavily dependent on coal-fired power plants, and many argue that the transition is not ready without adequate renewable energy and supporting infrastructure. Public opinion analysis is crucial in considering the introduction of electric vehicles in Indonesia due to the controversial nature of the transition. The opinion is transmitted through YouTube by taking comment data, then grouped into a topic to identify public opinion. The topic modeling method used is a BERTopic transformer model using IndoBERTweet in embedding. Once public opinion is modeled into a topic, changes in public opinion are evaluated using coherence score metrics and topic diversity as a measure of the consistency and diversity of the topic. The resulting topics have a coherence value of around 0.6 to 1 and a diversity value of 0.95838. This indicates that the resulting themes have strong semantic similarities and high diversity in terms of word usage and capture various aspects of text documents well.

Implementasi Model Hybrid CNN-LSTM untuk Optimasi Pengalaman Pengguna Perangkat Seluler

Yuhefizar — 2024-12-31

This research employs a convolutional neural network (CNN) with long short-term memory (LSTM) to analyse and predict the behaviour of users of mobile devices, utilising a dataset comprising 700 users. The model combines the strengths of convolutional neural networks (CNNs) in spatial feature extraction and long short-term memory (LSTM) networks in temporal sequential analysis. The results demonstrate that the model exhibits excellent performance, with 92% accuracy, 89% precision, 91% recall, and 90% F1 score. The temporal pattern analysis revealed significant variation between the user classes, with the intensive class showing consistently high usage, averaging 300 minutes per day. The key factors influencing the user experience were identified as app usage time (25%), screen on time (22%), and battery consumption (18%). The segmentation of users resulted in the identification of five distinct groups, with Segment 2 exhibiting the highest usage level (6.2 hours per day) and Segment 5 displaying the lowest (1.3 hours per day). The strong correlation (0.89) between app usage time and screen time serves to confirm the importance of optimising the performance of apps. These findings provide a basis for more effective service personalisation and more targeted app development, thereby paving the way for the optimisation of the user experience on mobile devices.

Manfaat Kecerdasan Buatan pada Proses Belajar Mengajar di Pendidikan Tinggi

Zen Munawar — 2024-12-31

This research aims to determine the benefits of using artificial intelligence in the teaching and learning process in higher education. The contribution of computer science, especially artificial intelligence, to the field of education is very much felt. From robotic teaching to the emergence of automated systems for answer sheet analysis, artificial intelligence always helps lecturers and students. This study also investigates the educational implications of emerging technologies on how students learn and how the teaching and learning process in higher education is developing. Data analysis is carried out comprehensively on various analytical developments applied worldwide such as computer science techniques applied to the education sector so as to summarize and highlight the role of artificial intelligence in the teaching and learning process. The latest technological advances and the increasing speed of application of new technologies in higher education are explored to predict the nature of higher education today and in a world where artificial intelligence becomes part of the higher education structure. The results of the study indicate several challenges for higher education institutions and student learning in adopting this technology for the teaching and learning process, support from students, and administration and the need to explore further research directions.