Edi Noersasongko

Sign language recognition is a critical component of assistive technologies for individuals with hearing and speech impairments. While vision-based approaches have shown promising performance, their reliability is often affected by illumination variations, occlusions, and background complexity. Wearable sensor–based solutions, particularly smart gloves integrating flex sensors and inertial measurement units (IMUs), provide a more stable alternative by directly capturing hand articulation and motion patterns. However, existing sensor-based methods frequently suffer from temporal instability, noise sensitivity, and limited discrimination among structurally similar gestures, which is especially challenging in Hijaiyah sign language, where many letters differ only by subtle finger configurations. This study proposes a robust real-time Multimodal Polynomial Fusion (MPF) framework for sensor-based sign language recognition using a flex–IMU smart glove, with a specific focus on Hijaiyah gestures as the application domain. The proposed framework applies nonlinear polynomial temporal smoothing within a sliding window to stabilize raw flex–IMU trajectories, followed by multimodal fusion to enhance gesture separability and temporal consistency. A large-scale multimodal dataset comprising 231,000 samples collected from 33 users performing 28 Hijaiyah gesture classes was constructed to enable rigorous subject-independent evaluation. Experimental results obtained from offline testing, session-aware analysis, and real-time streaming scenarios demonstrate that the proposed MPF framework consistently outperforms a baseline approach based on raw normalized signals. The proposed method improves recognition accuracy from 92.42% to 96.32%, while also achieving higher macro-level precision, recall, and F1-score. Furthermore, MPF significantly reduces misclassification rates and improves temporal stability, particularly for fine-grained Hijaiyah gestures with similar structural patterns. These results confirm that the proposed framework provides a robust and reliable solution for real-time wearable sign language recognition and offers practical benefits for Hijaiyah-based assistive communication systems.

Imbalanced dataset can cause obstacles to classification and result in a decrease in classification performance. There are several methods that can be used to deal the data imbalances, such as methods based on SMOTE and Generative Adversarial Networks (GAN). These methods are used for overcoming data oversampling so that the amount of minority data can increase and it can reach a balance with the majority data. In this research, the selected dataset is classified as a small imbalanced dataset of less than 200 records. The proposed method is the Gradually Generative Adversarial Network (GradGAN) model which aims to handle data imbalances gradually. The stages of the GradGAN model are adding the original minority dataset gradually so that it will create new minority datasets until a balance of data is created. Based on the algorithm flow described, the minority data is multiplied by the value of the variable that has been determined repeatedly to produce new balanced minority data. The test results on the classification of datasets from the GradGAN model produce an accuracy value of 8,3% when compare to that without GradGAN.

Video surveillance systems obtain a great interest as application-oriented studies that have been growing rapidly in the past decade. The most recent studies attempt to integrate computer vision, image processing, and artificial intelligence capabilities into video surveillance applications. Although there are so many achievements in the acquisition of datasets, methods, and frameworks published, there are not many papers that can provide a comprehensive picture of the current state of video surveillance system research. This paper provides a comprehensive and systematic review on the literature from various video surveillance system studies published from 2010 through 2019. Within a selected study extraction process, 220 journal-based publications were identified and analyzed to illustrate the research trends, datasets, methods, and frameworks used in the field of video surveillance, to provide an in-depth explanation about research trends that many topics raised by researchers as a focus in their researches, to provide references on public datasets that are often used by researchers as a comparison and a means of developing a test method, and to give accounts on the improvement and integration of network infrastructure design to meet the demand for multimedia data. In the end of this paper, several opportunities and challenges related to researches in the video surveillance system are mentioned.

Leprosy is one of the listed neglected tropical diseases which is continuing as a major health problem in Indonesia. Low adherence to Multi-Drug Therapy (MDT) against leprosy is still a significant obstacle in disease control, as the defaulters with incomplete cure remain as potential sources of infection, lead to patients' irreversible complications and multidrug resistance. This paper aims to describe the behavior of leprosy patients during the long-term medication of MDT drugs through mining association rule between the on-time of taking MDT to healthcare facilities using apriori algorithm. Data has been collected from manual leprosy registry cohort of 27 Primary Health Care (PHC) at Pekalongan District start from June, 1 ^st 2012 to June, 30 ^th 2014. This study converted the range date between the MDT collection attendances to the new label, which is called on-time attendance (OTA), and on time completion (OTC). Both of MB and PB patients shows the association between MDT 2, MDT 3 have a strong relation to the late of the completion treatment behavior of leprosy patients.

A news sentence is a sentence that tells or does something. In fact, The comparison between positive news and negative news broadcast on TV stations in Indonesia is 1:11. One point for positive newscasting and 11 points for negative newscasting. In the TVKU dataset because the original record number is too large, then the dataset used in this study only used 2.048 data. Which 1025 Positif and 1023 negative. Then obtained accurate results between labeling and testing. By using deep learning, we can get accurate results to analyze the sentiment from TVKU news.

Di Indonesia kepadatan arus lalu lintas terjadi pada jam berangkat dan pulang kantor, hari-hari libur panjang atau hari-hari besar nasional terutama saat hari raya Idul Fitri (lebaran). Mudik sudah menjadi tradisi bagi masyarakat Indonesia yang ditunggu-tunggu menjelang lebaran, berbondong-bondong untuk pulang ke kampung halaman untuk bertemu dan berkumpul dengan keluarga. Kegiatan rutin tahunan ini banyak di lakukan khususnya bagi masyarakat kota-kota besar seperti Jakarta, dimana diketahui bahwa Jakarta adalah Ibu kota negara Republik Indonesia dan menjadi tujuan merantau untuk mencari pekerjaan yang lebih layak yang merupakan harapan besar bagi masyarakat desa. Volume kendaraan bertambah sejak 7 hari menjelang lebaran sampai 7 hari setelah lebaran tiap tahunnya terutama pada arah keluar dan masuk wilayah Jawa Tengah yang banyak menjadi tujuan mudik. Volume kendaraan saat arus mudik yang selalu meningkat inilah yang akan diteliti lebih lanjut dengan metode ANFIS agar dapat menjadi alternatif solusi langkah apa yang akan dilakukan di tahun selanjutnya agar pelayanan lalu lintas, kemacetan panjang dan angka kecelakaan berkurang. Dengan input parameter ANFIS yang digunakan yaitu pengclusteran hingga 5 cluster, epoch 100, error goal 0 diperoleh performa terbaik ANFIS dengan K-Means clustering yang terbagi menjadi 3 cluster, epoch terbaik sebesar 20 dengan RMSE Training terbaik sebesar 0,1198, RMSE Testing terbaik sebesar 0,0282 dan waktu proses tersingkat sebesar 0,0695.Selanjutnya hasil prediksi diharapkan dapat bermanfaat menjadi alternatif solusi langkah apa yang akan dilakukan di tahun selanjutnya agar pelayanan lalu lintas lebih baik lagi. Kata kunci: angkutan lebaran, Jawa Tengah, ANFIS.

Speech recognition is becoming popular in current development on mobile devices. Presumably, mobile devices have limited computational power, memory size and battery life. In general, speech recognition is a heavy process that required large sample data within each window. Fast Fourier Transform (FFT) is the most popular transform in speech recognition. In addition, FFT operates in complex field with imaginary numbers. This paper proposes an approach based on discrete orthonormal Tchebichef polynomials as a possible alternative to FFT. Discrete Tchebichef Transform (DTT) shall be utilized here instead of FFT. The preliminary experimental result shows that speech recognition using DTT produces a simpler and efficient transformation for speech recognition. The frequency formants using FFT and DTT have been compared. The result showed that, they have produced relatively identical output in term of basic vowel and consonant recognition. DTT has the potential to provide simpler computing with DTT coefficient real numbers only than FFT on speech recognition.