In this study, we introduce an innovative approach that combines convolutional neural networks (CNN) with an attention mechanism (AM) to achieve precise emotion detection from speech data within the context of e-learning. Our primary objective is to leverage the strengths of deep learning through CNN and harness the focus-enhancing abilities of attention mechanisms. This fusion enables our model to pinpoint crucial features within the speech signal, significantly enhancing emotion classification performance. Our experimental results validate the efficacy of our approach, with the model achieving an impressive 90% accuracy rate in emotion recognition. In conclusion, our research introduces a cutting-edge method for emotion detection by synergizing CNN and an AM, with the potential to revolutionize various sectors.

The gravimetric method is one of the most accurate for determining soil water content (SWC). Several low-cost sensors have been developed to simplify measuring water content in soil by measuring soil moisture. However, the sensor must be calibrated to determine soil moisture parameters accurately. In this research, comparative analysis, and calibration of resistive and capacitive low-cost sensors were carried out. The calibration method for each sensor uses the gravimetric water content (GWC) and volumetric water content (VWC) methods. Measuring changes in SWC using sensors is performed in real time based on internet of things (IoT). Based on the measurements of the capacitive, resistive type 1, and resistive type 2 sensors with three repetitions, the linear regression R 2 values were obtained at 0.980, 0.827, and 0.942, respectively. Furthermore, a stability test is carried out to see how stable the sensor is when making measurements over a long period. The result is that the capacitive, resistive type 1, and resistive type 2 sensors have errors 1.971×10 -4 , 7.001×10 -4 , and 6.270×10 -4 . Based on the results obtained, capacitive sensors have the highest level of accuracy and stability. Furthermore, capacitive sensors are applied to IoT-based agriculture with long range (LoRa) as communication data.

Chemical sensors are increasingly used in healthcare because of their small size, durability, low resistance, and quick reaction time. This research aims to develop a pH biosensor utilizing an ion-sensitive field-effect transistor (ISFET) simulated with Silvaco technology computer-aided design (TCAD) software. The pH range of operation for the ISFET is 2 to 12. Sensitivity was evaluated based on the critical pH value and threshold voltage ( V th ) parameters across different gate channel lengths (250 nm, 200 nm, and 50 nm) and sensing membrane thicknesses (3 nm, 10 nm, and 20 nm). The sensitivity of different materials to pH levels was measured. Titanium dioxide (TiO 2 ) had the highest sensitivity at 57.98 mV/pH, followed by hafnium (IV) oxide (HfO 2 ) at 57.46 mV/pH, tantalum pentoxide (Ta 2 O 5 ) at 57.36 mV/pH, aluminium oxide (Al 2 O 3 ) at 55.05 mV/pH, and silicon nitride (Si 3 N 4 ) at 54.75 mV/pH. Notably, TiO 2 with a 200 nm gate channel length and a 3 nm sensing membrane thickness demonstrated the highest sensitivity. These findings highlight the potential of ISFETs, particularly those with TiO 2 sensing membranes, as robust and precise pH monitoring platforms in the biomedical industry.

Radiological imaging generates large data volumes for diagnosing. These data volumes still constitute an immense challenge for individual radiology practices, either when complying with the archive, hacker manipulation or swiftly distributing images among specialists as part of the diagnostic process. Analyzing images through illegal distortions may lead to inaccurate medical decisions. Watermarking techniques can be used to authenticate images, detect, and recover illegal changes made to teleradiology images as well. In this paper, a new blind medical image watermarking joint crypto-compression system is proposed to simultaneously achieve watermarking and compression. A discrete wavelet transform is applied to the medical image to locate different frequency sub-bands. The watermark is scrambled using one-dimensional chaotic map generator with uniform distribution modulo-one transformations for generating highly independent and uniformly distributed random chaotic sequences. A particle crypto-compression algorithm is applied for the insertion step: the scrambled watermark is inserted into the highest singular value derived from the compressed low-frequency sub-band, ensuring a delicate balance between imperceptibility and robustness. From the conducted tests and comparison of the obtained results with other techniques, we concluded that the robustness of the algorithm under various attacks was improved by the excellent value of the NC parameter, especially in the case of compression and geometric attacks.

Social networking is a place where humans can interact using the internet network to be able to disseminate information, discuss, exchange ideas, pour out their hearts, and share activities. Many social networks are popularly used, one of which is Twitter. Information can be received quickly using Twitter. In addition, various government agencies also use Twitter to be able to interact directly with the community so that every government policy is disseminated through this social network. Every government policy neglects to reap the pros and cons of society, both collectively and individually. As a result of the pros and cons, a trial called cyberbullying was recorded. Cyberbullying in various studies has been carried out to change a person's raw material so that with the application of information technology, identifying cyberbullying needs to be carried out further. The problem of cyberbullying is generally detected using the support vector machine (SVM) method. Cyberbullying detection is conducted in dealing with government policy data such as "cipta kerja" by using the SVM method which is optimized using the cubic kernel function. The accuracy value achieved in SVM uses a linear kernel function of 92.3% while using a cubic linear function of 90%.

The objectives of this research were: i) to study components of the internet of things (IoT) system to support community lime planting (CLP); in Maha Sarakham Province; ii) to develop IoT-CLP; and iii) to transfer IoT-CLP to the target community in Maha Sarakham Province. The research tools were: i) IoT-CLP system suitability assessment form; ii) IoT-CLP efficacy evaluation form; iii) a questionnaire to assess the effectiveness of the community lime planting; and iv) a survey of community satisfaction towards IoT-CLP. The research results showed that i) components of IoTCLP consist of: (1.1) data receivers: soil moisture, temperature, soil pH, fertilizer/nutrient; (1.2) control units: water supply, water spraying, soil pH control, and processing, fertilizer/nutrients; (1.3) operating systems: water supply, water spraying, soil pH communication and distribution system of fertilizer/soil nutrients water; (1.4) mobile application; ii) the effectiveness of IoT-CLP was evaluated at the highest level; and iii) the results of lime cultivation using IoT-CLP showed a higher yield than the traditional practices in terms of the amount of lime, size, and weight. The community was satisfied with IoT-CLP at the highest level.

Nowadays people are increasingly inclined to buy motorized vehicles because in addition to a light down payment, finance companies also provide convenience for the public in purchasing motorized vehicles. Even though the income level of the people in Indonesia is still relatively low, with a low down payment, the people are not too concerned about it. Honda showrooms carry out various forms of promotion and marketing so that the vehicles they sell get a response from consumers so they want to buy them. However, in fact there are still many forms of promotion that are not appropriate for consumers, so that in the motor vehicle loan process there are often obstacles caused by various factors. This study aims to create an analyst that can later be applied to computer systems, so that it can be said that by testing the system based on existing criteria, it will provide a definite answer in determining the creditworthiness of motorcycles to consumers.

Indoor modeling is a crucial aspect of architecture, engineering, and construction (AEC) projects. While terrestrial laser scanners (TLS) are the most popular tool for acquiring indoor geometry, processing point clouds from TLS scans with manual methods can be inefficient and error-prone. This study proposes a machine learning algorithm to automatically segment point clouds acquired by low-cost TLS. Random sample consensus (RANSAC), a simple yet effective algorithm for segmenting planar surfaces such as walls, ceilings, and floors, is used in the segmentation process. The resulting segmentation is then refined using density-based spatial clustering of application with noise (DBSCAN) to group nearby points that were not segmented correctly by RANSAC into the appropriate segment. The result is a segmented point cloud consisting of five indoor elements: wall, ceiling, floor, column, and interior. The algorithms were found to be effective for segmenting small and simple rooms. For larger or more complex rooms, segmentation can be performed by dividing the room into several parts and applying the algorithms to each partition. Overall, the study demonstrates the potential of machine learning algorithms for automating point cloud segmentation tasks in indoor modeling, especially for low-cost TLS scans.

Transportation activities require substantial costs: drivers, fuel, vehicle maintenance, vehicle and equipment procurement capital, and administrative activities. In the vehicle routing problem (VRP), all vehicles have the same load capacity. But in reality, the company has vehicles with different abilities. The heterogeneous vehicle routing problem (HVRP) is a variant of VRP. Transportation is a source of pollution. Therefore, electric vehicles are starting to replace fossil fuel vehicles. However, a few electric vehicle recharging stations remain, especially in Indonesia. Consequently, the authors build a heterogeneous vehicle routing problem model by considering the filling station using mixed integer programming. This research aims to create a model by determining distribution routes with minimum costs from several capacities of electric-powered freight fleets with different capabilities by considering refilling stations. Research contributions realize the transition towards sustainable energy, one of the priority issues in Indonesia's G20 presidency.

This study offers a 5G substrate integrated waveguide (SIW) antenna with vertical S zigzag shape slot. The suggested antenna's radiating patch depends on the shape of the slot. Two types of slots have been used, straight and zigzag S shape slot with vertical and horizontal direction, slots are etched on the patch to increase the antenna's overall bandwidth and gain. The suggested straight and zigzag SIW S slot antennas both resonate at 28 GHz, and the overall structure size is 7.10×14.93 mm. The presented design could achieve high gain, efficiency, and minimal losses, which are all important concerns. The presented antenna may produce a gain of 9.48 dB, a 95% efficiency, and a wider bandwidth of no less than 3.25 GHz at 28 GHz.