IAES International Journal of Artificial Intelligence (IJ-AI)
Not a member yet
1769 research outputs found
Sort by
Machine learning for the detection of soil pH, macronutrients, and micronutrients with crop and fertilizer recommendations
Get PDFThe study aims to determine the levels of soil parameters such as soil pH, macronutrients, and micronutrients. After determining said parameters, the system appropriately recommends crops and fertilizers suitable for the soil samples. For soil pH and macronutrient levels, i.e., nitrogen, phosphorus, and potassium, these parameters can be detected using the soil test kit. Meanwhile, for soil micronutrients, i.e., copper, iron, and zinc, there is a need for the development of appropriate assays for colorimetric processes that can be done for the appropriate determination of said micronutrients. Comparison of available machine learning such as support vector machine algorithm, naïve Bayes algorithms, and K-nearest neighbor algorithm is a must to determine the well-fit algorithm that is considered fast and has high predictive power in classification and regression. The outputs of the colorimetric and spectrometric processes are the inputs in the machine learning activities intended for crop and fertilizer recommendation
Comparison of faster region-based convolutional network for algorithms for grape leaves classification
Get PDFThe shapes of leaves distinguish the Indonesian grape variants. The grape leaves might look the same at first glance, but there are differences in leaf shapes and characteristics when observed closely. This research uses a deep learning method combined with the faster region-based convolutional neural network (R-CNN) algorithm with the Inception network architecture, ResNet V2, ResNet-152, ResNet-101, and ResNet-50, and uses COCO weights trained to classify five grape varieties through leaf images. The study collected 500 images to be used as an independent dataset. The results show that network improvements can effectively improve operating efficiency. There are also limitations to training scores because the F1 score value tends to stabilize or decrease at a certain point. In the Inception ResNet V2 architecture, with the highest average F1 score of 92%, the average computing time for training and testing is longer than other network architectures. This suggests that the algorithm can classify types of grapes based on their leaves
Averaged bars for cryptocurrency price forecasting across different horizons
Get PDFTechnical analysis uses past price movements and patterns to predict future trends and help traders make informed decisions about their cryptocurrency portfolios. This study investigates the effectiveness of different forecasting algorithms and features in predicting the future log return of cryptocurrency close price across various horizons. Specifically, we compare the performance of AdaBoost, light gradient boosting machine (LightGBM), random forest (RF), and k-nearest neighbor (KNN) regressors using Kline open, high, low, close (OHLC) prices data and averaged bars (Heikin-Ashi) features. Our analysis covers ten of the most capitalized cryptocurrencies: Cardano, Avalanche, Binance Coin, Bitcoin, Dogecoin, Polkadot, Ethereum, Solana, Tron, and Ripple. We have observed nuanced patterns in predictive performance across different cryptocurrencies, forecasting horizons and features. Then we have found that AdaBoost and RF models consistently exhibit a competitive performance, with LightGBM showing promising results for specific cryptocurrencies. The impact of forecast horizons on forecasting performance underscores the need for tailored forecasting models. In summary, the use of Kline OHLC data as features outperforms averaged bars in forecasting the first and second horizons, while averaged bars outperform Kline OHLC data for mid- to relatively long-term horizons (starting from the third horizon). Our findings suggest that averaged bars merit more attention from researchers instead of relying solely on Kline OHLC data
Enhancing precision medicine in neuroimaging: hybrid model for brain tumor analysis
Get PDFBrain tumors are a significant health challenge requiring precise diagnostic methods for optimal patient care. This study introduces a novel approach utilizing a convolutional neural network-based gated recurrent unit (CNN-GRU) for brain tumor detection. The method encompasses a rigorous preprocessing pipeline tailored for multi-modal magnetic resonance imaging (MRI) images, focusing on standardizing dimensions, normalizing pixel values, and enhancing contrast to facilitate robust tumor identification. Subsequently, temporal sequences of preprocessed images are analyzed by the CNN-GRU network to accurately pinpoint tumor regions. Evaluation on the BraTS2020 dataset, comprising diverse MRI scans with manual annotations, demonstrates the method's robust performance in tumor detection, reflecting real-world clinical complexities. Through meticulous preprocessing and model optimization, the approach achieves a remarkable accuracy rate of 99%, offering crucial insights for clinicians in treatment planning and prognosis prediction. Implemented using Python, the framework contributes to advancing brain tumor diagnosis and decision support systems, potentially enhancing personalized medicine and clinical practice. By improving diagnostic accuracy and patient outcomes, this research underscores the importance of integrating advanced computational techniques with medical imaging to address critical healthcare challenges effectively
A hybrid framework for wild animal classification using fine-tuned DenseNet121 and machine learning classifiers
Get PDFOver the past few decades, wildlife monitoring has become an active research area. Various topics like animal-vehicle collision, human-animal conflict, animal poaching, population demography, and tracking of animal behaviour come under wildlife monitoring. Different methods have been used for wild animal monitoring, out of which machine learning (ML) and deep learning (DL) are widely used for automatic detection and classification of species from digital images. Both ML and DL have their advantages and disadvantages. A hybrid model has been proposed by integrating the advantage of DL and ML to detect and classify animals automatically. Publicly available datasets of five wild animals are used to train the model, and for testing the model, a dataset is created by capturing the images with the help of a camera and mobile device in different locations and under various environmental conditions. Two approaches are proposed using a hybrid model: a pre-trained dense convolution neural network 121 (DenseNet121) model is used in the first approach, and a finetuned DenseNet121 model is used in the second approach for feature extraction. Extracted features from the pre-trained and finetuned DenseNet121 model are fed into ML classifiers such as extreme gradient boosting (XGBoost), random forest (RF), and naïve Bayes (NB) for classification. When the performance was analysed, the second approach performed better than the first
Artificial intelligence applications in agriculture: a systematic review of literature
Get PDFArtificial intelligence (AI) is transforming agriculture by offering innovative solutions to persistent challenges. This systematic literature review explores the most studied AI applications in agriculture, emphasizing crop management, agronomic decision-making, early detection of diseases and pests, and climate change adaptation. Using the preferred reporting items for systematic reviews and meta-analyses (PRISMA) methodology, 700 publications were retrieved from databases such as Scopus, ScienceDirect, and IEEE Xplore, with 104 relevant articles selected after applying strict inclusion and exclusion criteria. The findings underscore the importance of machine learning and image processing in tailoring agronomic practices to specific plot conditions and microclimates. These tools enable early identification and control of plant diseases and pests, reducing crop losses and dependence on chemicals. Nonetheless, challenges remain, particularly regarding accessibility for smallholder farmers, high implementation costs, and limited data infrastructure. While AI offers significant potential to enhance agricultural productivity, sustainability, and resilience, addressing these limitations is crucial. A balanced, inclusive approach is essential to ensure AI’s benefits are widely distributed and contribute to long-term food security and environmental sustainability
Optimizing traffic lights at unbalanced intersections using deep reinforcement learning
Get PDFUnbalanced intersectional traffic flow increases vehicle delays, fuel consumption, and pollution. This study investigates the application of deep reinforcement learning (DRL) to optimize traffic signal timing at the Pamelisan intersection in Denpasar, Indonesia. Real-world traffic data were incorporated into a SUMO microsimulation environment to train DRL agents using the deep Q-network (DQN) algorithm. Experimental results show that DRL-based optimization reduced the average vehicle waiting time from 594.49 seconds (static control) to 169.44 seconds and 173.10 seconds for agents trained without and with noise, respectively. The average vehicle speed remained stable at 5.6–5.97 m/s across all scenarios, indicating enhanced traffic efficiency without adverse effects. The findings underscore the effectiveness and adaptability of DRL in addressing traffic inefficiencies, optimizing them, and offering a robust solution for dynamic traffic management at unbalanced traffic intersections in urban areas
Myoelectric grip force prediction using deep learning for hand robot
Get PDFArtificial intelligence (AI) has been widely applied in the medical world. One such application is a hand-driven robot based on user intention prediction. The purpose of this research is to control the grip strength of a robot based on the user’s intention by predicting the grip strength of the user using deep learning and electromyographic signals. The grip strength of the target hand is obtained from a handgrip dynamometer paired with electromyographic signals as training data. We evaluated a convolutional neural network (CNN) with two different architectures. The input to CNN was the root mean square (RMS) and mean absolute value (MAV). The grip strength of the hand dynamometer was used as a reference value for a low-level controller for the robotic hand. The experimental results show that CNN succeeded in predicting hand grip strength and controlling grip strength with a root mean square error (RMSE) of 2.35 N using the RMS feature. A comparison with a state-of-the-art regression method also shows that a CNN can better predict the grip strength
Learning assistance module based on a small language model
Get PDFThis paper presents the development of a low-cost learning assistant embedded in an NVIDIA Jetson Xavier board that uses speech and gesture recognition, together with a long language model for offline work. Using the large language model (LLM) Phi-3 Mini (3.8B) model and the Whisper (model base) model for automatic speech recognition, a learning assistant is obtained under a compact and efficient design based on extensive language model architectures that give a general answer set of a topic. Average processing times of 0.108 seconds per character, a speech transcription efficiency of 94.75%, an average accuracy of 9.5/10 and 8.5/10 in the consistency of the responses generated by the learning assistant, a full recognition of the hand raising gesture when done for at least 2 seconds, even without fully extending the fingers, were obtained. The prototype is based on the design of a graphical interface capable of responding to voice commands and generating dynamic interactions in response to the user's gesture detection, representing a significant advance towards the creation of comprehensive and accessible human-machine interface solutions
Design and analysis of reinforcement learning models for automated penetration testing
Get PDFOur paper proposes a framework to automate penetration testing by utilizing reinforcement learning (RL) capabilities. The framework aims to identify and prioritize vulnerable paths within a network by dynamically learning and adapting strategies for vulnerability assessment by acquiring the network data obtained from a comprehensive network scanner. The study evaluates three RL algorithms: deep Q-network (DQN), deep deterministic policy gradient (DDPG), and asynchronous episodic deep deterministic policy gradient (AE-DDPG) in order to compare their effectiveness for this task. DQN uses a learned model of the environment to make decisions and is hence called model-based RL, while DDPG and AE-DDPG learn directly from interactions with the network environment and are called model-free RL. By dynamically adapting its strategies, the framework can identify and focus on the most critical vulnerabilities within the network infrastructure. Our work is to check how well the RL technique picked security vulnerabilities. The identified vulnerable paths are tested using Metasploit, which also confirmed the accuracy of the RL approach's results. The tabulated findings show that RL promises to automate penetration testing tasks