AIAmil Ahmad Ilham

Amil Ahmad Ilham

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Amil Ahmad Ilham has a total of 54 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Improving the Performance of Voice Lie Detection Using Mel Frequency Cepstral Coefficients and Long Short Term Memory Models

    by Dewi Kusumawati, Amil Ahmad Ilham, Andani Achmad, Ingrid Nurtanio

    The objective of this study is to show that the combination of the Mel frequency cepstral coefficient (MFCC) and long short-term memory (LSTM) can be an effective approach for voice lie detection. To improve the performance of voice lie detection, a modified MFCC was used to extract important features in voice. An MFCC was modified by adding zero-crossing rate, audio entropy, and energy entropy parameters to detect changes in tone in each voice frame. LSTM was used to detect and classify voice-based lies. Datasets were obtained from the video recordings of the trial of a suspect. A total of 847 voice datasets were obtained after applying the time stretching augmentation technique where the audio duration was changed from 28.0 s to 4 s per video. The lie classification process was performed using the LSTM method that was equipped with additional dropout and dense layers and optimized using the adaptive moment estimation (Adam) optimizer. The results showed that the combination of the MFCC and LSTM achieved a classification accuracy level of 97% and an area under the curve value of 0.97 using epoch parameters of 200, Adam optimizer, and learning rate of 0.0001. This study concluded that the addition of zero-crossing rate, audio entropy, and energy entropy parameters to the MFCC extraction feature and the use of Adam optimizer in LSTM improved the accuracy of voice lie detection.

  • Open Access

    Enhancing Relational Database Efficiency through Algorithmic Query Tuning in Virtual Memory Systems

    by Nurlina Yulis, Amil Ahmad Ilham, Andani Achmad, Faizal Arya Samman

    The rapid evolution of virtual memory-based relational database systems has significantly advanced data processing capabilities. However, the efficiency of these systems largely depends on query execution optimization, which can be enhanced through algorithmic query tuning techniques. This study investigates the impact of these techniques on enhancing query performance in virtual memory-based relational databases. Various algorithmic methods were analyzed to optimize query execution plans, with a focus on key performance indicators such as execution time, CPU and memory usage, disk I/O, and cache hit ratio. The systematic application of these methods revealed effective strategies for performance enhancement. Results show substantial improvements in execution time, resource utilization, and scalability. This work offers valuable insights for database administrators and system architects, highlighting the role of algorithmic query tuning in managing the growing demands for data processing. Future research endeavors should explore the realm of AI-driven automation, with a particular focus on enhancing query optimization techniques. Additionally, there is a pressing need to investigate advanced security measures that safeguard data integrity within expansive, large-scale systems. By adopting innovative approaches, we can ensure robust protection and efficient performance in an increasingly data-driven world.

  • Open Access

    Attention-Driven Image Captioning for Mobile Accessibility of the Visually Impaired

    by Dessy Santi, Amil Ahmad Ilham, Syafaruddin, Ingrid Nurtanio

    In a world increasingly reliant on visual information, individuals with visual impairments face significant challenges in understanding their environment. This paper introduces an attention-based image captioning model to improve accessibility for visually impaired users. The model integrates ResNet-152 for visual feature extraction, long short-term memory (LSTM) for text processing, and an attention mechanism to generate contextual image descriptions. Captured images are processed via a mobile device, then the description text is translated into Bahasa and converted to speech in real-time using text-to-speech technology. The system shows an average inference time of 2.99 seconds per image, enabling real-time use. The model is tested on the Flickr dataset and new datasets covering a variety of environments and object interactions. Experimental results show superior performance on the Flickr dataset (bilingual evaluation understudy (BLEU)-1: 0.59, metric for evaluation of translation with explicit ordering (METEOR): 0.25). Performance on real-world datasets is slightly lower, indicating challenges in generalizing to scenarios with occluded objects and inconsistent text. Future research will focus on scaling up real-world datasets, adversarial training, and integrating the system into devices such as smart glasses or canes for wider accessibility.

  • Open Access

    Transformers for aerial images semantic segmentation of natural disaster-impacted areas in natural disaster assessment

    by Deny Wiria Nugraha, Amil Ahmad Ilham, Andani Achmad, Ardiaty Arief

    Aerial image segmentation of natural disaster-impacted areas and detailed and automatic natural disaster assessment are the main focus of this study. Detecting and recognizing objects on aerial images of areas impacted by natural disasters and assessing natural disaster-impacted areas are still difficult problems. To solve these problems, this study utilizes four of the latest transformer-based semantic segmentation network models, bidirectional encoder representation from image transformers (BEIT), dense prediction transformer (DPT), OneFormer, and SegFormer, and proposes a detailed and automatic natural disaster assessment of the segmented image. The SegFormer model achieved the first-best result, and the OneFormer model achieved the second-best result. The SegFormer model outperformed OneFormer by 1.58% higher for the mean accuracy value and 4.28% for the mean intersection over union (mIoU) value. All receiver operating characteristics (ROC) curves have mean area under curve (AUC) values above 0.9, which means that the SegFormer model performs well in generating semantic segmentation images. The fuzzy c-means (FCM) clustering algorithm performed well and could automatically cluster the natural disaster assessments into four categories. This study has produced semantic segmentation of aerial images of areas impacted by natural disasters and natural disaster assessments, which can be used in natural disaster management systems.


  • Open Access

    Convolutional Neural Network with Feature Extraction to Improve the Classification Accuracy of Multi-Class Facial Skin Disorders

    by Rismayani, Amil Ahmad Ilham, Andani Achmad, Muhammad Rifqy Yudhiestra Rachman

    This study aims to improve the accuracy of multi-class facial skin disorder classification using a convolutional neural network (CNN) enhanced with feature extraction. The CNN method for classifying multi-class facial skin disorders uses color feature extraction using color moment (CM) and Laplacian of Gaussian (LoG) for direct shape with image data. Multi-class facial skin disorders include oily, hyperpigmentation, acne, redness, blackhead, and normal. A public dataset is used with 7151 images with a balanced number of data classes. Researchers divided the data set into 80% for training and 20% for testing. Experiments are carried out through training and testing with 100 epochs, resulting in an accuracy of 85% for CNN, 66% for the CM-CNN, 80% for LoG-CNN, and 91% for CM-LoG-CNN. The highest classification accuracy is achieved with the CM-LoG-CNN combination.


  • Open Access

    Optimization of Herbal Plant Classification Using Hybrid Method Particle Swarm Optimization With Support Vector Machine

    by Amriana Amriana, Amil Ahmad Ilham, Andani Achmad, Yusran Yusran

    The classification process applied in this study helps identify the many kinds of herbal plants. Herbal plant leaf features are used based on color, shape, and texture. Particle Swarm Optimization and Support Vector Machine (PSO-SVM) hybridization are applied in the classification process to increase classification and identification accuracy. A well-liked metaheuristic approach for solving optimization issues is Particle Swarm Optimization (PSO). Particles look around the search area for the best responses. A particle swarm is initially initialized randomly within the search area via the PSO algorithm. Every particle's mobility is determined by both its own experience and the experiences of the other particles in the swarm. Each particle keeps track of the best solution it has ever found and the swarm's most extraordinary remedy that has so far been discovered. The Hybrid approach concurrently selects features for the SVM and optimizes its parameters. The kernel function's gamma value non-linearly maps an input space to a high-dimensional feature space. At the same time, the C parameter determines the trade-off between fitting error minimization and model complexity. The Gaussian kernel parameter is set to determine the optimal parameter value of the RBF kernel function. Feature selection solves the issue by eliminating redundant, associated, and irrelevant features. A confusion matrix is utilized in the evaluation to gauge the system's performance. The results demonstrated an improvement in accuracy, with the hybrid PSO-SVM using test data achieving an accuracy of 98% compared to the SVM method, achieving a 91% accuracy.

  • Open Access

    Performance Analysis of Feature Mel Frequency Cepstral Coefficient and Short Time Fourier Transform Input for Lie Detection using Convolutional Neural Network

    by Dewi Kusumawati, Amil Ahmad Ilham, Andani Achmad, Ingrid Nurtanio

    This study aims to determine which model is more effective in detecting lies between models with Mel Frequency Cepstral Coefficient (MFCC) and Short Time Fourier Transform (STFT) processes using Convolutional Neural Network (CNN). MFCC and STFT processes are based on digital voice data from video recordings that have been given lie or truth information regarding certain situations. Data is then pre-processed and trained on CNN. The results of model performance evaluation with hyper-tuning parameters and random search implementation show that using MFCC as Voice data processing provides better performance with higher accuracy than using the STFT process. The best parameters from MFCC are obtained with filter convolutional=64, kerneconvolutional1=5, filterconvolutional2=112, kernel convolutional2=3, filter convolutional3=32, kernelconvolutional3 =5, dense1=96, optimizer=RMSProp, learning rate=0.001 which achieves an accuracy of 97.13%, with an AUC value of 0.97. Using the STFT, the best parameters are obtained with filter convolutional1=96, kernel convolutional1=5, convolutional2 filters=48, convolutional2 kernels=5, convolutional3 filters=96, convolutional3 kernels=5, dense1=128, Optimizer=Adaddelta, learning rate=0.001, which achieves an accuracy of 95.39% with an AUC value of 0.95. Prosodics are used to compare the performance of MFCC and STFT. The result is that prosodic has a low accuracy of 68%. The analysis shows that using MFCC as the process of sound extraction with the CNN model produces the best performance for cases of lie detection using audio. It can be optimized for further research by combining CNN architectural models such as ResNet, AlexNet, and other architectures to obtain new models and improve lie detection accuracy.


  • Open Access

    Applying LSTM and GRU Methods to Recognize and Interpret Hand Gestures, Poses, and Face-Based Sign Language in Real Time

    by Amil Ahmad Ilham, Ingrid Nurtanio, Ridwang, Syafaruddin

    This research uses a real-time, human-computer interaction application to examine sign language recognition. This work develops a rule-based hand gesture approach for Indonesian sign language in order to interpret some words using a combination of hand movements, mimics, and poses. The main objective in this study is the recognition of sign language that is based on hand movements made in front of the body with one or two hands, movements which may involve switching between the left and right hand or may be combined with mimics and poses. To overcome this problem, a research framework is developed by coordinating hand gestures with poses and mimics to create features by using holistic MediaPipe. To train and test data in real time, the long short time memory (LSTM) and gated recurrent unit (GRU) approaches are used. The research findings presented in this paper show that hand gestures in real-time interactions are reliably recognized, and some words are interpreted with the high accuracy rates of 94% and 96% for the LSTM and GRU methods, respectively.


  • Open Access

    Peningkatan Efisiensi Monitoring Status Gizi Anak melalui Pembangunan dan Pendampingan Penggunaan Sistem Informasi untuk Kader Posyandu

    by Mukarramah Yusuf, Amil Ahmad Ilham, Ady Wahyudi Paundu, Elly Warni, A Sungkurawira Batara U, Paula C Chudori, Amiqatun Nasyati Yusri, Zulfiqry Saadputra

    Posyandu volunteers (cadre) have an essential role in the early detection of stunting in children under two years, achieved through routine weighing and monitoring of children's growth. This community service activity helps Posyandu volunteers in their role of recording and reporting. Posyandu volunteers use the SIP Book (Posyandu Information System) to record and report the results of activities at Posyandu and monitor maternal and child health, including children's nutritional status. Manually recording the nutritional status of children registered in Posyandu by the volunteers creates time inefficiencies in organizing the Posyandu. For this reason, an information system was built to precisely meet the needs of Posyandu volunteers in their role to monitor stunting in their work area. This information system is website-based and manages nutritional status data for toddlers at Posyandu. This system was tested on Posyandu volunteers in Nisombalia village, Maros Regency, South Sulawesi. Evaluation through questionnaire shows an improvement (from bad to very good) in terms of ease of recording and monitoring the nutritional status of children registered with Posyandu.


  • Open Access

    Aerial imagery segmentation of natural disaster-affected areas using deep convolutional networks for disaster assessment

    by Deny Wiria Nugraha, Amil Ahmad Ilham, Andani Achmad, Ardiaty Arief

    After a natural disaster, aerial imagery from an unmanned aerial vehicle (UAV) is an invaluable resource for impact assessment and disaster management. Traditio After a natural disaster, aerial imagery from an unmanned aerial vehicle (UAV) is an invaluable resource for impact assessment and disaster management. Traditio