|
|
Direction-of-arrival estimation for conventional
co-prime arrays using probabilistic Bayesian neural
networks
Wael Elshennawy, Orange Business Services, Egypt
Abstract:
The paper investigates the direction-of-arrival (DOA) estimation of narrow
band signals with conventional co-prime arrays by using efficient
probabilistic Bayesian neural networks (PBNN). A super resolution DOA
estimation method based on Bayesian neural networks and a spatially
overcomplete array output formulation overcomes the pre-assumption
dependencies of the model-driven DOA estimation methods. The proposed
DOA estimation method utilizes a PBNN model to capture both
data and model uncertainty. The developed PBNN model is trained to
do the mapping from the pseudo-spectrum to the super resolution spectrum.
This learning-based method enhances the generalization of untrained
scenarios, and it provides robustness to non-ideal conditions, e.g.,
small angle separation, data scarcity, and imperfect arrays, etc. Simulation
results demonstrate the root mean square error (RMSE) and loss
curves of the PBNN model in comparison with deterministic model and
spatial-smoothing MUSIC (SS-MUSIC) method. The proposed Bayesian
estimator improves the DOA estimation performance for the case of low
signal-to-noise ratio (SNR) or with a limited number of model trainable
variables or spatially adjacent signals.
Full Paper (in PDF)
|
Hybrid Deep Learning for Assembly Action Recognition in Smart Manufacturing
Abdul Matin, University of Technology Sydney, Australia
Md Rafiqul Islam, University of Technology Sydney, Australia
Yeqian Zhu, University of Technology Sydney, Australia
Xianzhi Wang, University of Technology Sydney, Australia
Huan Huo, University of Technology Sydney, Australia
Guandong Xu, University of Technology Sydney, Australia
Abstract:
Deep learning algorithms have become essential in assembly action recognition
(AAR) for driving advancements in intelligent manufacturing.
While numerous sensor systems and algorithms are developing, their
real-world applicability and robustness within the manufacturing sector
need validation. Artificial intelligence (AI) applications in manufacturing
have gained significant attraction in both academic and industrial
circles. One key aspect of future smart manufacturing is identifying the
actions of manufacturing workers, particularly monitoring repetitive assembly
tasks, to guide them and improve efficiency. This recognition
facilitates real-time efficiency measurement and evaluation of workers
while providing augmented reality instructions to enhance their performance
on the job. This paper introduces a hybrid deep-learning approach
combining 3D CNN and ConvLSTM2D models to monitor assembly
tasks to recognize human actions within the manufacturing context.
The model’s performance is evaluated through simulations conducted
on the HA4M dataset, comprising diverse multimodal data-capturing
actions executed by various individuals constructing an epicyclic gear
train (EGT). The proposed hybrid model demonstrated superior performance
on the HA4M dataset relative to baselines.
Full Paper (in PDF)
|
Automated Fracture Detection from Pelvic X-ray:
The Impact of Appropriate Labeling on the
Performance of Deep Convolutional Neural Network
Rashedur Rahman, University of Hyogo, Japan
Naomi Yagi, University of Hyogo, Japan
Keigo Hayashi, Hyogo Prefectural Harima-Himeji General Medical Center, Japan
Akihiro Maruo, Hyogo Prefectural Harima-Himeji General Medical Center, Japan
Hirotsugu Muratsu, Hyogo Prefectural Harima-Himeji General Medical Center, Japan
Sayoji Kobashi, University of Hyogo, Japan
Abstract:
Pelvic X-rays (PXRs) are essential diagnostic tools used to visualize the
pelvic region and assess pelvic fractures. The rising incidence of pelvic
fractures leads to increased radiologist workload and initial misdiagnoses.
As a result, there is a growing need for automated tools to assist doctors
in pelvic fracture detection. Artificial intelligence has advanced
recently, resulting in several methods for diagnosing PXRs for fractures.
However, concerns regarding annotation accuracy and the limitations of
PXRs due to constrained viewing angles persist. Some fractures are only
visible in 3D computed tomography (CT) images, and it is difficult to
understand their visibility in PXR. This study proposes a method for
using annotations from pelvic CT to label PXRs, focusing on fracture
visibility. Additionally, the impact of labeling PXRs based on visibility
to fracture detection performance in PXR images is examined. First,
all fractures in CT images are annotated using a 3D surface annotation
approach. Next, annotated pseudo PXRs are synthesized from CT
images utilizing digitally reconstructed radiographs (DRRs). The annotated
pseudo PXRs serve as references for accurately labeling fractures
in corresponding PXRs. By training a Resnet-101-based deep convolutional
neural network (DCNN) with the labeled datasets considering
fracture visibility, the proposed method significantly improved fracture
detection performance, achieving an Area Under the Receiver Operating
Characteristic (AUROC) of 0.9114. The AUROC of the conventional
annotation method was 0.8202.
Full Paper (in PDF)
|
Exploring Oversampling Techniques for Fraud
Detection with Imbalanced Classes
Sultan Alharbi, University of Technology Sydney, Australia
Abdulrhman Alorini, University of Technology Sydney, Australia
Khaled Alahmadi, University of Technology Sydney, Australia
Hadeel Alhosaini, University of Technology Sydney, Australia
Yeqian Zhu, University of Technology Sydney, Australia
Xianzhi Wang, University of Technology Sydney, Australia
Abstract:
Each year, credit card fraud has caused significant losses for financial institutions
and individuals worldwide. Financial institutions must detect
credit card fraud to prevent customers from being charged for products
they did not order. Class imbalance has been a standing challenge for
credit card transactions, as the number of fraudulent transactions is significantly
lower than that of non-fraudulent transactions. In this paper,
we comprehensively evaluate five oversampling techniques, namely Synthetic
Minority Oversampling Technique (SMOTE), Adaptive Synthetic
Sampling (ADASYN), Borderline SMOTE, Random Oversampling, and
SMOTE Support Vector Machine (SMOTE SVM), in combination with
seven machine learning techniques (namely XGBoost, Random Forest,
K-Nearest Neighbor, Naive Bayes, Support Vector Machine, LightGBM,
and Convolution Neural Network). Our results show oversampling generally
improves fraud detection performance and SMOTE SVM is the
better oversampling method than other methods under test. Notably,
it achieved an accuracy of 76.47% when used with KNN on the smaller
dataset and 99.93% with CNN on the larger dataset used in our experiments.
Full Paper (in PDF)
|
Generation of Clothing Items with Jamdani Motif
Elements Using Automated Generative Adversarial
Networks
Hujaifa Islam, Samiur Rahman Abir, Md. Sakibur Rahman, Hasan Mahmud, Mohammad Shafiul Alam,
Ahsanullah University of Science and Technology, Bangladesh
Abstract:
Clothing serves as an artistic medium for humans to express their preferences,
thoughts, and cultural heritage, while the application of machine
learning, particularly Generative Adversarial Networks (GANs),
remains largely unexplored in the realm of clothing production and design,
with designers currently relying on their imaginative skills to create
diverse styles. In this article, Conditional Generative Adversarial Networks
(cGAN) are used to suggest an automated approach. Neural style
transfer and cGAN algorithms are employed. to create traditional clothing
with distinctive patterns and a variety of styles. For this study,
the Fashion MNIST and Jamdani Motif Dataset datasets were both employed.
The conditional GAN model was used to produce several styles
of apparel using the MNIST dataset. The Neural Style Transfer model
is then used to combine the created picture with the Jamdani Motif pattern
from the Jamdani Motif dataset. Using Otsu's image segmentation
technique, the foreground, and background of the resulting picture are
separated. Performance scores of this model are as follows: Inception
Score is 1.3573909, Frechet inception distance is 1272.222597, Kernel Inception
Distance is 636200.667, Coverage Metric is 33.79799. We polled
several people on our work output, and the results are detailed in a later
section. Generate Jamdani clothing using single pattern and remove extra
regions using image segmentation.
Full Paper (in PDF)
|
In-Depth Analysis of Automated Acne Disease
Recognition and Classification
Afsana Ahsan Jeny,
Masum Shah Junayed,
University of Connecticut, Storrs, USA
Md Robel Mia,
Daffodil International University, Bangladesh
Md Baharul Islam
Florida Gulf Coast University, USA
Abstract:
Facial acne is a common disease, especially among adolescents, negatively affecting individuals both physically and psychologically. Classifying acne is vital for providing appropriate treatment. Traditional visual inspection or expert scanning is time-consuming and challenging to differentiate acne types. This paper introduces an automated expert system for acne recognition and classification. The proposed method employs a machine learning-based technique to classify and evaluate six types of acne diseases, facilitating the diagnosis process for dermatologists. The preprocessing phase includes contrast improvement, smoothing filter application, and RGB to Lab color conversion to eliminate noise and improve classification accuracy. Next, a clustering-based segmentation method, k-means clustering, is applied to segment the disease-affected regions, which then proceed to the feature extraction step. Characteristics of these disease-affected regions are extracted using a combination of gray-level co-occurrence matrix (GLCM) and statistical features. Finally, five different machine learning classifiers are employed to classify acne diseases.
The experimental results show that Random Forest (RF) achieves the highest accuracy of 98.50%, which is promising compared to state-of-the-art methods.
Full Paper (in PDF)
|
A Real-Time Anti-Aliasing Approach for 3D
Applications Using Deep Convolutional Neural
Network
F. M. Jamius Siam, Zahidul Islam Prince, Ahmed Nafisul Bari,
BRAC University, Bangladesh
Jia Uddin,
Woosong University, South Korea
Abstract:
In real-time 3D applications, delivering smooth edges in the output images is essential, mainly due to limitations in resolution, memory, and processing power. This paper proposes a deep convolutional neural network-based model designed to address this aliasing issue. Aliasing in an image is characterized by hard, jagged edges that are present especially when the edges do not line up with the pixel grid of the output device. Our approach leverages a deep convolutional neural network to learn these jagged patterns in images from a training dataset and generates anti-aliased output images. The model's architecture includes several layers of convolutional neural networks, max-pooling layers, and convolutional transpose layers. During the experimental analysis, we used a dataset comprising demo 3D scenes created with both the Unity and Unreal game engines. This dataset contains raw and super-sampled images along with images processed with various other anti-aliasing techniques. To assess performance, we used both SSIM and PSNR scores as metrics to analyze the model’s accuracy. The experimental results show that our proposed model not only competes with but often surpasses other state-of-the-art methods like MSAA, FXAA, TAA, and SMAA, by achieving higher SSIM and PSNR scores.
Full Paper (in PDF)
|
Investigation of Emotional Effects on Brain Network
Stimulation through EEG Signals
Mahfuza Akter Maria, M. A. H. Akhand,
Khulna University of Engineering and Technology, Bangladesh
Md Abdus Samad Kamal,
Gunma University, Japan
Abstract:
There has been growing evidence in recent years which supports that different brain areas are involved in processing emotions. As a result, research on emotion from the perspective of brain networks is becoming popular. The connectivity strength of this network can be changed with different mental states, which can be identified through different frequency bands of the brain signal. In this study, brain functional and effective connectivity networks have been constructed from DEAP emotional EEG data to study how emotion influences patterns of this connectivity. According to the investigation results, more direct correlations are found under positive emotions than negative ones. The brain regions operate more synchronously, and there is less directed flow of information between brain regions during negative emotions. The correlation between brain regions, whether direct or inverse, is higher in the lower frequency band than in the higher frequency band. The flow of information from one brain region to another brain region increases with higher frequency, and there is more synchrony between brain regions in the Gamma frequency band. The findings of this study have substantial implications for the practical application of EEG-based emotion analysis, as well as prospective avenues for future research in this field.
Full Paper (in PDF)
|
A Low-cost IoT-based Meteorological System Using
LoRaWAN and Embedded Technologies:
Architecture and Future Trends
Norbert Dajnowski, Andrew Guest,
York St John University, UK
Aminu Bello Usman,
University of Sunderland, UK
Abdulrazaq Abba,
University of East London, UK
Saifur Rahman Sabuj,
BRAC University, Bangladesh
Abstract:
The field of meteorological station development is undergoing continuous advancements, driven by the pursuit of more precise data acquisition while also maintaining cost-effectiveness. To achieve this objective, governments and businesses are increasingly harnessing Internet of Things (IoT) platforms to deploy hyperlocal and highly sophisticated meteorological stations. These stations are designed to offer real-time analysis of weather conditions and forecasts with unparalleled precision. In this study, we have designed and implemented a robust yet affordable meteorological system capable of collecting various weather parameters. This system is integrated with a low-cost, long-range data transmission technology, enabling multiple nodes to access the internet through the LoRaWAN network server. Additionally, we have developed a user-friendly graphical user interface (GUI) application for visualizing meteorological data. Our proposed solution demonstrated impressive capabilities. The system consistently recorded and transmitted essential meteorological parameters, such as temperature, humidity, pressure, and wind speed, with high accuracy and reliability. Additionally, our GUI application facilitated user-friendly access to this data, offering clear visual representations of weather conditions and station performance.
Full Paper (in PDF)
|
The Integrity of Source Code Commenting:
Benchmark Dataset and Empirical Analysis
Maksuda Islam, Md Safayat Hossen, Ahsanul Haque, Md. Nazmul Haque, Lutfun Nahar Lota,
Islamic University of Technology, Bangladesh
Abstract:
Code comments are a vital software feature for program cognition & software maintainability. For a long time, researchers have been trying to find ways to ensure the consistency of code-comment.
While doing that, two of the raised problems have been dataset scarcity and language dependency.
To address both problems in this paper, we created a dataset using C# projects; there are no annotated datasets yet on C#. 9,310 code-comment pairs of different C# projects
were extracted from a data pool. 4,922 code-comment pairs were annotated after removing NULL, constructor, and variable. Both method-comment and class-comment were considered in this study.
We employed two evaluation metrics for the dataset, one is Krippendorff’s Alpha which showed 95.67% similarity among the rating of three annotators for all the pairs & other is Bilingual Evaluation Understudy (BLEU) to validate our human-curated dataset.
An ensemble machine learning model with topic modeling is also proposed, which obtained 96.2% using the performance metric AUC-ROC after fitting the model to our proposed dataset.
Full Paper (in PDF)
|
|
