Vietnam Journal of Computer Science

In intelligent information systems data play a critical role. The issue of missing data is one of the commonplace problems occurring in data collected in the real world. The problem stems directly from the very nature of data collection. In this paper, the notion of handling missing values in a real-world application of computational intelligence is considered. Two experimental campaigns were conducted, evaluating different approaches to the missing values imputation on Random Forest-based classifiers, trained using modern cybersecurity benchmarks datasets: CICIDS2017 and IoT-23. In result of the experiments it transpired that the chosen algorithm for data imputation has a severe impact on the results of the classifier used for network intrusion detection. It also comes to light that one of the most popular approaches to handling missing data — complete case analysis — should never be used in cybersecurity.

Traffic forecasting has emerged as an important task for developing intelligent transportation systems. Recent works focus on representing traffic as graph operation and using graph neural networks for spatial–temporal prediction. Most of the approaches assume a predefined graph structure based on node distances. However, spatial dependencies change over time in many scenarios of traffic flow. In this regard, this study takes an investigation capturing the spatial and temporal dependencies with no prior knowledge structure of traffic road networks. Specifically, we propose a multi-step prediction model named Dynamic Spatial Transformer WaveNet Network (DSTWN) to capture the dynamic conditions and directions of traffic flow in which a temporal convolution layer is adopted for the long time sequence and a spatial transformer layer is proposed to capture the dynamic spatial dependencies. Furthermore, we introduce a new traffic dataset, which is collected from the vehicle detection system in an urban area (UVDS). In particular, compared with existing benchmark traffic data, UVDS contains more complicated spatial information, which is similar to many real-world scenarios of traffic flow. Experiments on both benchmark traffic datasets indicate the promising results of DSTWN compared with state-of-the-art models in this research field.

Bird species identification is a relevant and time-consuming task for ornithologists and ecologists. With growing amounts of audio-annotated data, automatic bird classification using machine learning techniques is an important trend in the scientific community. Analyzing bird behavior and population trends helps detect other organisms in the environment and is an important problem in ecology. Bird populations react quickly to environmental changes, which make their real-time counting and tracking challenging and very useful. A reliable methodology that automatically identifies bird species from audio would therefore be a valuable tool for the experts in different scientific and applicational domains. The goal of this work is to propose a methodology to identify bird sounds. In this paper, we explore deep learning techniques that are being used in this domain, such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) to classify the data. In deep learning, audio problems are commonly approached by converting them into images using audio feature extraction techniques such as Mel Spectrograms and Mel Frequency Cepstral Coefficients (MFCCs). We propose and test multiple deep learning and feature extraction combinations in order to find the most suitable approach to this problem.

Medical X-rays are one of the primary choices for diagnosis because of their potential to disclose previously undetected pathologic changes, non-invasive qualities, radiation dosage, and cost concerns. There are several advantages to creating computer-aided detection (CAD) technologies for X-Ray analysis. With the advancement of technology, researchers have lately used the deep learning approach to obtain high accuracy outcomes in the CAD system. With CAD, computer output may be utilized as a backup option for radiologists, assisting doctors in making the best selections. Chest X-Rays (CXRs) are commonly used to diagnose heart and lung problems. Automatically recognizing these problems with high accuracy might considerably improve real-world diagnosis processes. However, the lack of standard publicly available datasets and benchmark research makes comparing and establishing the best detection algorithms challenging. In order to overcome these difficulties, we have used the VinDr-CXR dataset, which is one of the latest public datasets including 18,000 expert-annotated images labeled into 22 local position-specific abnormalities and 6 globally suspected diseases. To improve the identification of chest abnormalities, we proposed a data preparation procedure and a novel model based on YOLOv5 and ResNet50. YOLOv5 is the most recent YOLO series, and it is more adaptable than previous one-stage detection algorithms. In our paper, the role of YOLOv5 is to locate the abnormality location. On the other side, we employ ResNet for classification, avoiding gradient explosion concerns in deep learning. Then we filter the YOLOv5 and ResNet results. The YOLOv5 detection result is updated if ResNet determines that the image is not anomalous.

In this paper, we present an unified framework for understanding hand action from the first-person video. The proposed framework composes two main components. The first component estimates three-dimensional (3D) hand joints from RGB images. Two network structures derived from the baseline HopeNet network are proposed: convolutional neural networks (CNNs) which are traditional multi-layer CNN and CNN combining with GraphCNN to perform 3D hand pose estimation, without the use of GraphUNet as in baseline HopeNet method. The second component of the framework recognizes hand action from skeleton stream. We first deploy two recent advanced neuronal networks that are PA-ResGCN and Double-feature Double-motion (DDNet). To focus more on the hand pose changes, we improve DDNet with two normalization strategies of the hand joints. Finally, we fuse PA-ResGCN with our improved DDNet to still boost the recognition performance. We evaluate our proposed methods on First-Person Hand Action Benchmark dataset. Experiments show that our model for 3D hand joints estimation achieves the best precision (36.6 mm). Our hand joint normalization strategies improve the original DDNet from 0.71% to 4.05% of accuracy with the ground-truth hand pose while the improvement is significantly larger (from 2.96% to 10.98%) with the estimated hand pose. The late fusion schemes outperform different state-of-the-art methods for the hand action recognition with the highest accuracy of 86.67%. These experimental results show potential and extendable possibilities for developing practical first-person vision applications.

Designing new materials with desired properties is a complex and time-consuming process. One of the most challenging factors of the design process is the huge search space of possible materials. Machine learning methods such as $k$-nearest neighbors, support vector machine (SVM) and artificial neural network (ANN) can contribute to this process by predicting materials properties accurately. Properties of multi-principal element alloys (MPEAs) highly depend on alloys’ phase. Thus, accurate prediction of the alloy’s phase is important to narrow down the search space. In this paper, we propose a solution of employing SVM method with hyperparameters tuning and the use of weighted values for prediction of the alloy’s phase. Using the dataset consisting of the experimental results of 118 MPEAs, our solution achieves a cross-validation accuracy of 90.2%. We confirm the superiority of this score over the performance of ANN statistically. On the other dataset containing 401 MPEAs, our SVM model is comparable to ANN and exhibits 70.6% cross-validation accuracy. We also found that additional variables, including average melting temperature and standard deviation of melting temperature, increase prediction accuracy by 3.34% in the best case.