Search

This paper proposes an adaptive sliding mode control (ASMC) with a modified reduced-order proportional integral observer (MROPIO) to provide speed control for a permanent magnet synchronous motor against unknown disturbances, such as system parameter uncertainties and external load torque variation. The ASMC method, based on a new sliding mode reaching law, improves chattering in the control signal and reduces the time required for the system trajectory to reach the sliding mode surface. The MROPIO estimates unknown disturbances and provides feedforward compensation for the ASMC, thus overcoming incorrect estimations of unknown disturbances during speed jumps can cause overshoots in the speed response. All the proposed methods were implemented on an industrial motor drive. The experimental results confirm the validity of the proposed method for practical applications.

Visual impairment community, especially blind people have a thirst for assistance from advanced technologies for understanding and answering the image. Through the development and intersection between vision and language, Visual Question Answering (VQA) is to predict an answer from a textual question on an image. It is essential and ideal to help blind people with capturing the image and answering their questions automatically. Traditional approaches often utilize the strength of convolution and recurrent networks, which requires a great effort for learning and optimizing. A key challenge in VQA is finding an effective way to extract and combine textual and visual features. To take advantage of previous knowledge in different domains, we propose BERT-RG, the delicate integration of pre-trained models into feature extractors, which relies on the interaction between residual and global features in the image and linguistic features in the question. Moreover, our architecture integrates a stacked attention mechanism that exploits the relationship between textual and visual objects. Specifically, the partial regions of images interact with partial keywords in question to enhance the text-vision representation. Besides, we also propose a novel perspective by considering a specific question type in VQA. Our proposal is significantly meaningful enough to develop a specialized system instead of putting forth the effort to dig for unlimited and unrealistic approaches. Experiments on VizWiz-VQA, a practical benchmark dataset, show that our proposed model outperforms existing models on the VizWiz VQA dataset in the Yes/No question type.

This article proposes a method to improve the accuracy of a vernier absolute magnetic encoder. The encoder consists of a master and a nonius multipolar magnetic track. Sinusoidal signals from the master and nonius tracks are used to infer the absolute information. Unfortunately, these signals are contaminated by nonideal factors such as different amplitudes, dc-offsets, phase shifts, and random noise. Moreover, harmonics existing in the encoder signals distort the vernier principle and significantly affect the accuracy of the encoder. To address these problems, the present article proposes an efficient method with three main parts. The first is an observer phase-locked loop (OPLL), which is used to estimate the phase and eliminate the nonideal factors. The second is nonlinear phase compensation, which is used to correct the vernier principle that deviated due to the existing harmonics. Finally, a pole pitch compensation method is introduced to modulate the master phase angle from the OPLL to eliminate the harmonic distortion. The proposed method can eliminate the nonideal factors, harmonic distortion and improve the accuracy of the encoder. All the proposed methods were implemented on an ARM STM32F407ZG. The experimental results confirm the validity of the proposed method for practical applications.

The main objective of this study is to calibrate Discrete Element Modeling (DEM) input parameters for Vietnamese DT84 variety soybeans. For this purpose, the shape of the soybeans was firstly analy...

Natural language processing techniques contribute more and more in analyzing legal documents recently, which supports the implementation of laws and rules using computers. Previous approaches in representing a legal sentence often based on logical patterns that illustrate the relations between concepts in the sentence, often consist of multiple words. Those representations cause the lack of semantic information at the word level. In our work, we aim to tackle such shortcomings by representing legal texts in the form of abstract meaning representation (AMR), a graph-based semantic representation that gains lots of polarity in NLP community recently. We present our study in AMR Parsing (producing AMR from natural language) and AMR-to-text Generation (producing natural language from AMR) specifically for legal domain. We also introduce JCivilCode, a human-annotated legal AMR dataset which was created and verified by a group of linguistic and legal experts. We conduct an empirical evaluation of various approaches in parsing and generating AMR on our own dataset and show the current challenges. Based on our observation, we propose our domain adaptation method applying in the training phase and decoding phase of a neural AMR-to-text generation model. Our method improves the quality of text generated from AMR graph compared to the baseline model. (This work is extended from our two previous papers: “An Empirical Evaluation of AMR Parsing for Legal Documents”, published in the Twelfth International Workshop on Juris-informatics (JURISIN) 2018; and “Legal Text Generation from Abstract Meaning Representation”, published in the 32nd International Conference on Legal Knowledge and Information Systems (JURIX) 2019.).

Falls are one of the most serious dangers for elderly people who live alone at home. It has become a widespread issue all across the world. Reliable fall detection systems can help to mitigate the negative consequences of accidentally falling. Several techniques for automatically falling detection machines have been suggested in the past. The existing technologies are classified into three types of fall detectors: wearable sensor-based, ambient device-based, and computer vision-based approaches. This paper focuses on a dataset comprising signals from wearable sensors, ambient sensors, and vision devices. We propose a novel feature subset selection to reduce the number of effective input attributes based on a hybridized metaheuristic - an Adaptive Particle Swarm and Grey Wolf Optimization (APGWO). Classification results use various machine learning classifiers such as Logistic Regression (LR), K-Nearest Neighbor, Support Vector Machine (SVM), Decision Tree (DT), Naïve Bayes (NB), Random Forest (RF), and Multilayer Perceptron (MLP), show that the proposed approach is highly effective. Classification accuracy and F1 score can reach as high as 99% and 96%, respectively.

Diabetes leads to health problems for hundreds of millions of people globally every year. Available medical records of patients quantify symptoms, body features, and clinical laboratory test values, which can be used to perform biostatistics analysis aimed at finding patterns or features undetectable by current practice. In this work, we proposed a machine learning model to predict the early onset of diabetes patients. It is a novel wrapper-based feature selection utilizing Grey Wolf Optimization (GWO) and an Adaptive Particle Swam Optimization (APSO) to optimize the Multilayer Perceptron (MLP) to reduce the number of required input attributes. Moreover, we also compared the results achieved using this method and several conventional machine learning algorithms approaches such as Support Vector Machine (SVM), Decision Tree (DT), K-Nearest Neighbor (KNN), Naive Bayesian Classifier (NBC), Random Forest Classifier (RFC), Logistic Regression (LR). Computational results of our proposed method show not only that much fewer features are needed, but also higher prediction accuracy can be achieved (96% for GWO - MLP and 97% for APGWO - MLP). This work has the potential to be applicable to clinical practice and become a supporting tool for doctors/physicians.

The optimal operation for different states such as normal and contingency cases of a power system has a very important role in the operation. Therefore, it is necessary to analyze contingencies in the system so as the most severe cases should be considered for integrating into the optimal power flow (OPF) problem and the security-constrained optimal power flow (SCOPF) becomes an important problem for considering in the power system operation. This paper proposes a combined pseudo-gradient based particle swarm optimization with constriction factor (PGPSO) and the differential evolution (DE) method for solving the SCOPF problem. The PGPSO-DE method is a newly developed method for utilizing the advantages of the pseudogradient guided PSO method with a constriction factor and the DE method. The proposed PGPSO-DE has been tested on the IEEE 30 bus system for the normal case and the contingency case with two types of the objective function. The results yielded from the proposed method have been validated via comparing to those from the conventional PSO, DE, and other methods reported in the literature. The comparisons for the results obtained from the proposed PGPSODE method have shown that it is very effective to solve the large-scale and complex SCOPF problem.

A precision matrix is the inverse of a covariance matrix. In this paper, we study the problem of estimating the precision matrix with a known graphical structure under high-dimensional settings. We propose a simple estimator of the precision matrix based on the connection between the known graphical structure and the precision matrix. We obtain the rates of convergence of the proposed estimators and derive the asymptotic normality of the proposed estimator in the high-dimensional setting when the data dimension grows with the sample size. Numerical simulations are conducted to demonstrate the performance of the proposed method. We also show that the proposed method outperforms some existing methods that do not utilize the graphical structure information.

Energy plays a very important role in the development of a country in many aspects of economic, social, environmental to security and defense. Correct forecasting of energy demand will make an important contribution to the implementation of energy, socio-economic and environmental policies and ensure the sustainable development of the country. Therefore, the selection of an appropriate energy forecasting model will play an important role in setting appropriate strategies and policies in the future. This article will synthesize energy forecasting models in the world, in-depth exploration of the POLES model and consider its applicability in energy forecasting in Vietnam.

If a perforation occurs as a result of a flexible endoscopic procedure, suturing through urgent laparoscopy or open surgery may be required to repair the perforation because suturing is normally stronger than closure using existing endoscopic devices. Suturing with stitches and knots, widely adopted in open or laparoscopic surgery, is still not possible in flexible endoscopy. This is because of the confined space of the natural orifice and target area, high levels of motion dexterity and force needed for stitching and knot-tying, and critical size and strength requirements of wound closure. We present a novel flexible endoscopic robotic suturing system that is able to suture gastrointestinal defects without opening up the patient’s body like in open or laparoscopic surgery. This system features a robotic needle driver and a robotic grasper, both of which are flexible, through-the-scope (small in sizes), and dexterous with five degrees of freedom. The needle driver, facilitated by the grasper, enables the surgeon to control a needle through teleoperation to make stitches and knots in flexible endoscopy. Successful in vivo trials were conducted in the rectum of a live pig to confirm the feasibility of endoscopic suturing and knot-tying using the system in a realistic surgical scenario (not possible with existing devices which are all manually driven). This new technology will change the way how surgeons close gastrointestinal defects. NRF (Natl Research Foundation, S’pore) Accepted version

To improve the accuracy of permanent magnet (PM) rotor position estimation for a hybrid stepper motor (HSM), the authors propose a sensorless control design based on the back electromagnetic force (back-EMF) caused by magnetic induction and harmonic rejection via an orthogonal third-order phase-locked loop (PLL 3 rd) and an integral harmonic filter (IHF). The accuracy of estimation is analysed considering PM rotor position estimation errors. The harmonics of the back-EMF signal are eliminated by the IHF before entering the PLL 3 rd, which synchronises and decreases position estimation error. This technique is a simple, reliable, and effective method for implementing sensorless control of an HSM. An industrial test-bench was used to implement the proposed scheme. The experimental results are presented to validate the effectiveness of the proposed estimation method.

Crash occurrence is a complex phenomenon, and crashes associated with pedestrians and bicyclists are even more complex. Furthermore, pedestrian- and bicyclist-involved crashes are typically not reported in detail in state or national crash databases. To address this issue, developers created the Pedestrian and Bicycle Crash Analysis Tool (PBCAT). However, it is labour-intensive to manually identify the types of pedestrian and bicycle crash from crash-narrative reports and to classify different crash attributes from the textual content of police reports. Therefore, there is a need for a supporting tool that can assist practitioners in using PBCAT more efficiently and accurately. The objective of this study is to develop a framework for applying machine-learning models to classify crash types from unstructured textual content. In this study, the research team collected pedestrian crash-typing data from two locations in Texas. The XGBoost model was found to be the best classifier. The high prediction power of the XGBoost classifiers indicates that this machine-learning technique was able to classify pedestrian crash types with the highest accuracy rate (up to 77% for training data and 72% for test data). The findings demonstrate that advanced machine-learning models can extract underlying patterns and trends of crash mechanisms. This provides the basis for applying machine-learning techniques in addressing the crash typing issues associated with non-motorist crashes.

Rainfall–runoff modelling is complicated due to numerous complex interactions and feedback in the water cycle among precipitation and evapotranspiration processes, and also geophysical characteristics. Consequently, the lack of geophysical characteristics such as soil properties leads to difficulties in developing physical and analytical models when traditional statistical methods cannot simulate rainfall–runoff accurately. Machine learning techniques with data-driven methods, which can capture the nonlinear relationship between prediction and predictors, have been rapidly developed in the last decades and have many applications in the field of water resources. This study attempts to develop a novel 1D convolutional neural network (CNN), a deep learning technique, with a ReLU activation function for rainfall–runoff modelling. The modelling paradigm includes applying two convolutional filters in parallel to separate time series, which allows for the fast processing of data and the exploitation of the correlation structure between the multivariate time series. The developed modelling framework is evaluated with measured data at Chau Doc and Can Tho hydro-meteorological stations in the Vietnamese Mekong Delta. The proposed model results are compared with simulations of long short-term memory (LSTM) and traditional models. Both CNN and LSTM have better performance than the traditional models, and the statistical performance of the CNN model is slightly better than the LSTM results. We demonstrate that the convolutional network is suitable for regression-type problems and can effectively learn dependencies in and between the series without the need for a long historical time series, is a time-efficient and easy to implement alternative to recurrent-type networks and tends to outperform linear and recurrent models.

We present our method for tackling a legal case retrieval task by introducing our method of encoding documents by summarizing them into continuous vector space via our phrase scoring framework utilizing deep neural networks. On the other hand, we explore the benefits from combining lexical features and latent features generated with neural networks. Our experiments show that lexical features and latent features generated with neural networks complement each other to improve the retrieval system performance. Furthermore, our experimental results suggest the importance of case summarization in different aspects: using provided summaries and performing encoded summarization. Our approach achieved F1 of 65.6% and 57.6% on the experimental datasets of legal case retrieval tasks.

We investigate a robust single machine scheduling-location problem with uncertainty in edge lengths. Jobs are located at the vertices of a given tree. Given a location for a single machine, the jobs travel to the location and are processed there sequentially. The goal is to find a location of the machine and simultaneously a sequence to minimize the makespan value in the worst-case. We use the concept of gamma-robustness to model uncertainty. Our main result is a polynomial time algorithm.

Long Short Term Memory (LSTM) and Convolutional Neural Network (CNN) are efficiently applied to natural language processing, especially sentiment analysis. CNN employs filters to capture local dependencies while LSTM designs a cell to memorize long-distance information. However, integrating these advantages into one model is challenging because of overfitting in training. To avoid this problem, we propose a freezing technique to learn sentiment-specific vectors from CNN and LSTM. This technique is efficient for integrating the advantages of various deep learning models. We also observe that semantically clustering documents into groups is more beneficial for ensemble methods. According to the experiments, our method achieves competitive results on the five well-known datasets: Pang & Lee movie reviews, Stanford Twitter Sentiment and Stanford Sentiment Treebank for sentence level, IMDB large movie reviews, and SenTube for document level.

In the evaluation of magnetic field data, edge enhancement and detection techniques are important treatments for the interpretation of geological structures. In general geological sense, contiguity of deep and shallow magnetic sources leads to weak and intense anomalies that complicates the interpretation to disclose adjacent anomalous sources. Many of the existing filters for edge detection in magnetics mostly have the disadvantage that they require a reduction to pole transformation as the pre-process of the data or they cannot balance weak and intense anomalies and therefore fail in detecting edges of deep and shallow sources simultaneously. This study presents an improved edge detection filter LAS (logistic function of the analytical signal), based on the generalised logistic function configured by the ratio of derivatives of the analytical signal. This novel approach has the capability of reducing the dependence on the direction of the magnetization and also balancing anomalies of sources at different levels of depth. The feasibility of the method is examined on both theoretical and real data cases comparatively with some other methods that utilize the analytical signal in their basis. In comparison, the results demonstrate that the LAS method provides more accurate estimation of edge localization.

Work Zone (WZ) related injuries and fatalities from vehicular crashes are rising because of the increasing maintenance needs of the aging transportation infrastructure in the United States. Although many potential benefits from commercial WZ intrusion technologies have been recognized, they have not seen widespread adoption due to cost, simplicity, and limitations in their design to only provide post-intrusion alarms. Hence, there is significant interest and motivation in developing Smart Work Zone (SWZ) safety systems that predictively warn workers of vehicle intrusion threats with more accuracy and greater lead times. In this paper, we review the perception and alarm requirements, and propose the emerging Light Detection and Ranging (LiDAR) as a suitable sensing component. Specifically, we analyze LiDAR resolution and frequency in the context of a WZ application, detail an initial vehicle detection and tracking algorithm running realtime between sensor updates, and provide annotated datasets for a stationary roadside LiDAR in highway / urban road environments.

The rise of big data and big compute has brought modern neural networks to many walks of digital life, thanks to the relative ease of constructing large models that scale to the real world. Current successes of Transformers and self-supervised pretraining on massive data have led some to believe that deep neural networks will be able to do almost everything once we have sufficient data and computational resources. However, neural networks are fast to exploit surface statistics but fail miserably to generalize to novel combinations. This is because they are not designed for deliberate reasoning -- the capacity to deliberately deduce new knowledge out of the contextualized data. This tutorial reviews recent developments to extend the capacity of neural networks to "learning-to-reason'' from data, where the task is to determine if the data entails a conclusion. This capacity opens up new ways to generate insights from data through arbitrary compositional querying without the need of predefining a narrow set of tasks. The tutorial consists of four parts. The first part covers the learning-to-reason framework, and explains how neural networks can serve as a strong backbone for reasoning through its natural operations such as binding, attention & dynamic computational graphs. The second part goes into more detail on how neural networks perform reasoning over unstructured and structured data, and across modalities. The third part reviews neural memories and their role in reasoning. The last part discusses generalization to novel combinations, under less supervision and with more knowledge.

Predicting the interaction between a compound and a target is crucial for rapid drug repurposing. Deep learning has been successfully applied in drug-target affinity (DTA) problem. However, previous deep learning-based methods ignore modeling the direct interactions between drug and protein residues. This would lead to inaccurate learning of target representation which may change due to the drug binding effects. In addition, previous DTA methods learn protein representation solely based on a small number of protein sequences in DTA datasets while neglecting the use of proteins outside of the DTA datasets. We propose GEFA (Graph Early Fusion Affinity), a novel graph-in-graph neural network with attention mechanism to address the changes in target representation because of the binding effects. Specifically, a drug is modeled as a graph of atoms, which then serves as a node in a larger graph of residues-drug complex. The resulting model is an expressive deep nested graph neural network. We also use pre-trained protein representation powered by the recent effort of learning contextualized protein representation. The experiments are conducted under different settings to evaluate scenarios such as novel drugs or targets. The results demonstrate the effectiveness of the pre-trained protein embedding and the advantages our GEFA in modeling the nested graph for drug-target interaction.

This paper identifies a model of software evolution that is prevalent in large, long-lived academic research tool suites (3L-ARTS). This model results in an "archipelago" of related but haphazardly organized architectural "islands", and inherently induces technical debt. We illustrate the archipelago model with examples from two 3L-ARTS archipelagos identified in literature.

With internet video gaining increasing popularity and soaring to dominate network traffic, extensive studies are being carried out on how to achieve higher Quality of Experience (QoE) with the delivery of video content. Associated with the chunk-based streaming protocol, Adaptive Bitrate (ABR) algorithms have recently emerged to cope with the diverse and fluctuating network conditions by dynamically adjusting bitrates for future chunks. This inevitably involves predicting the future throughput of a video session. Some of the session features like Internet Service Provider (ISP), geographical location, and so on, could affect network conditions and contain helpful information for this throughput prediction. In this article, we consider how our knowledge about the session features can be utilized to improve ABR quality via customized parameter settings. We present our ABR-independent, QoE-driven, feature-based partition method to classify the logged video sessions so that different parameter settings could be adopted in different situations to reach better quality. A variation of Decision Tree is developed for the classification and has been applied to a sample ABR for evaluation. The experiment shows that our approach can improve the average bitrate of the sample ABR by 36.1% without causing the increase of the rebuffering ratio where 99% of the sessions can get improvement. It can also improve the rebuffering ratio by 87.7% without causing the decrease of the average bitrate, where, among those sessions involved in rebuffering, 82% receives improvement and 18% remains the same.

In this paper, we propose an advanced approach to improve the accuracy of stepper motors based on an effective phase compensator. The proposed approach includes an adaptive current controller and an adaptive position controller. The proposed adaptive feed-forward proportional-resonant (AFPR) current control technique with a neural network (NN) is an improvement over the conventional proportional-integral algorithm. In which, the NN algorithm is applied to obtain the optimal parameters for the AFPR controller. This results in an enhancement of the stepper motor current tracking. In addition, the advanced position controller is a combination of a damping and a phase-compensated (D&PC) method to reduce steady-state error and variations during low-speed operation. The position tracking performance is further appreciated by applying an advanced proportional integral derivative (PID) controller during high-speed operation. A bandpass filter is designed to switch the motor between low- and high-speed operational regimes. The proposed approach does not require any changes in the structure of the hardware. Therefore, it can be widely applied to a diverse set of industrial applications. The performance of the proposed approach was validated by conducting experiments under practical conditions.

Due to increasing penetration of wind power plants (WPP), power grids are facing significant power quality challenges at the point of interconnection (POI). To achieve stable and robust power system operation, reactive power plays a vital role. Reactive power needed to compensate voltage fluctuations at the POI of the WPP varies with a short-circuit ratio (SCR). The reactive power capability of a WPP is also limited and largely depends on various operating conditions. In a physical system, it is difficult to find the correlation among critical parameters for voltage control. In this paper, a data-driven voltage control approach is proposed for a grid-connected WPP. Two regression models are developed through surface fitting using MATLAB curve fitting toolbox: One model based on simulation data is to determine the required reactive power for grid voltage compensation; another model based on field measurement data is to determine the reactive power characteristics of the WPP. The reactive power compensation device is capacitors in this study. Two controllers, a central WPP controller, and a capacitor controller are designed, and their effectiveness is validated through several case and sensitivity studies.

One way to increase storage density is using a shingled magnetic recording (SMR) disk. We propose a novel use of SMR disks with RAID (redundant array of independent disks) arrays, specifically building upon and compared with a basic RAID 4 arrangement. The proposed scheme (called RAID 4SMR) has the potential to improve the performance of a traditional RAID 4 array with SMR disks. Our evaluation shows that compared with the standard RAID 4, when using update in-place in RAID arrays, RAID 4SMR with garbage collection not just can allow the adoption of SMR disks with a reduced performance penalty, but offers a performance improvement of up to 56%.

In the present study, Rotation Forest ensemble was integrated with different base classifiers to develop different hybrid models namely Rotation Forest based Support Vector Machines (RFSVM), Rotati...

The aim of object-oriented domain-driven design (DDD) is to iteratively develop software around a realistic domain model. Recent work in DDD use an annotation-based extension of object-oriented programming language to build the domain model. This model becomes the basis for a ubiquitous language and is used as input to generate software. However, the annotation-based extensions of these work do not adequately address the primitive and essential structural and behavioural modelling requirements of practical software. Further, they do not precisely characterise the software that is generated from the domain model. In this paper, we propose a DSL-based DDD method to address these limitations. We make four contributions: (1) an annotation-based domain-specific language (DSL) named DCSL , whose annotation extension expresses a set of essential structural constraints and the essential behaviour of a domain class. (2) A structural mapping between the state and behaviour spaces of a domain class. This mapping enables a technique for generating the behavioural specification. (3) A technique that uses DCSL to support behavioural modelling with UML activity diagram. (4) A 4-property characterisation of the software generated from the domain model. We demonstrate our method with a Java software tool and evaluate DCSL in the context of DDD.

Demand Response program is applied in many countries as an effective instrument to regulate the electricity consumption. In this program, time of use (TOU) tariff is used widely. Optimal TOU pricing according to different objectives was mentioned in this paper such as peak load reduction, improving load curve, energy conservation, avoiding a new peak load. This is a problem with multiobjective functions in different unit of measurement and is solved by PSO algorithm. An example to find optimal TOU tariff for one utility is also presented in this paper.

During the COVID-19 pandemic, many countries implemented international travel restrictions that aimed to contain viral spread while still allowing necessary cross-border travel for social and economic reasons. The relative effectiveness of these approaches for controlling the pandemic has gone largely unstudied. Here we developed a flexible network meta-population model to compare the effectiveness of international travel policies, with a focus on evaluating the benefit of policy coordination. Because country-level epidemiological parameters are unknown, they need to be estimated from data; we accomplished this using approximate Bayesian computation, given the nature of our complex stochastic disease transmission model. Based on simulation and theoretical insights we find that, under our proposed policy, international airline travel may resume up to 58% of the pre-pandemic level with pandemic control comparable to that of a complete shutdown of all airline travel. Our results demonstrate that global coordination is necessary to allow for maximum travel with minimum effect on viral spread.

A problem of ultra-high frequency radiofrequency identification (UHF RFID) reader is the reading range because one antenna is used for transmitting carrier and backscattered signal. In this study, a carrier-cancellation circuit is used with a commercial directional coupler to obtain a good isolation between transmitter (Tx) and receiver (Rx) by suppressing the transmitting leakage signal. The suppression level of the circuit can be obtained by applying the Genetic Algorithm (GA) to optimize values of digitally tunable capacitors. Experimental result of the S24 of the directional coupler is approximately of -70 dB which showed the good isolation between Tx and Rx at 900 MHz.

This paper presents CovRelex, a scientific paper retrieval system targeting entities and relations via relation extraction on COVID-19 scientific papers. This work aims at building a system supporting users efficiently in acquiring knowledge across a huge number of COVID-19 scientific papers published rapidly. Our system can be accessed via https://www.jaist.ac.jp/is/labs/nguyen-lab/systems/covrelex/., Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations,Online,EACL

Architectural decay imposes real costs in terms of developer effort, system correctness, and performance. Over time, those problems are likely to be revealed as explicit implementation issues (defects, feature changes, etc.). Recent empirical studies have demonstrated that there is a significant correlation between architectural "smells" -- manifestations of architectural decay -- and implementation issues. In this paper, we take a step further in exploring this phenomenon. We analyze the available development data from 10 open-source software systems and show that information regarding current architectural decay in these systems can be used to build models that accurately predict future issue-proneness and change-proneness of the systems' implementations. As a less intuitive result, we also show that, in cases where historical data for a system is unavailable, such data from other, unrelated systems can provide reasonably accurate issue- and change-proneness prediction capabilities.

Cardiovascular diseases kill approximately millions of people globally every year. Heart failure (HF) occurs when the heart cannot pump enough blood to meet the needs of the body. This means that the heart is overworked and unable to respond to the speed and demands of other activities. This will lead to fatigue and shortness of breath while performing daily activities. In this research, the authors apply a machine learning model to predict heart failure patients’ survival based on the original set of available medical features. Therefore, a novel wrapper-based feature selection utilizing an Adaptive Particle Swarm Grey Wolf Optimization (APSGWO) is proposed to enhance the architecture of Multilayer Perceptron (MLP) and reduce the number of required input attributes. Moreover, we also compared the results of our proposed method and several conventional machine learning models such as Support Vector Machine (SVM), Decision Tree (DT), K–Nearest Neighbor, Naive Bayesian Classifier (NBC), Random Forest (RF), and Logistic Regression (LR). The results of our method show not only that much fewer features are needed, but also higher accuracy can be accomplished, 81% for Adaptive Particle Swarm Grey Wolf Optimization—Multilayer Perceptron (APSGWO—MLP). This work can be applied in practice to become an effective tool to support the diagnosis for doctors.

Video question answering (Video QA) presents a powerful testbed for human-like intelligent behaviors. The task demands new capabilities to integrate video processing, language understanding, binding abstract linguistic concepts to concrete visual artifacts, and deliberative reasoning over spacetime. Neural networks offer a promising approach to reach this potential through learning from examples rather than handcrafting features and rules. However, neural networks are predominantly feature-based - they map data to unstructured vectorial representation and thus can fall into the trap of exploiting shortcuts through surface statistics instead of true systematic reasoning seen in symbolic systems. To tackle this issue, we advocate for object-centric representation as a basis for constructing spatio-temporal structures from videos, essentially bridging the semantic gap between low-level pattern recognition and high-level symbolic algebra. To this end, we propose a new query-guided representation framework to turn a video into an evolving relational graph of objects, whose features and interactions are dynamically and conditionally inferred. The object lives are then summarized into resumes, lending naturally for deliberative relational reasoning that produces an answer to the query. The framework is evaluated on major Video QA datasets, demonstrating clear benefits of the object-centric approach to video reasoning., Comment: Accepted by IJCNN 2021

The objective of this study is to control rotational angles of a 3-degree freedom robot for tracking to the reference trajectories. That is a parallel robot, named Delta, with complex movements in shaping, processing with high efficiency, high load capacity, widely used in industry. The PID controller has been successfully developed for the Delta robot. However, when changing the robot’s parameters such as load, input coupling and friction, the PID controller is difficult to archive control criteria. This article proposes and tests a solution to improve the PID controller by combining it with a recurrent fuzzy neural network (RFNN) controller, so-called RFNN-PID controller. In the proposed solution, the PID controller plays the main role of controlling the Delta robot and the RFNN controller takes charge of a supplemental role to gain with the changes of control conditions. The RFNN-PID and PID controllers will be tested in the same conditions in MATLAB/Simulink. Simulations illustrate that the proposed controller is better than the traditional one, obtaining a response time of about 3.9 ± 0.1 (s) without steady-state error.

Video Question Answering (Video QA) is a powerful testbed to develop new AI capabilities. This task necessitates learning to reason about objects, relations, and events across visual and linguistic domains in space-time. High-level reasoning demands lifting from associative visual pattern recognition to symbol-like manipulation over objects, their behavior and interactions. Toward reaching this goal we propose an object-oriented reasoning approach in that video is abstracted as a dynamic stream of interacting objects. At each stage of the video event flow, these objects interact with each other, and their interactions are reasoned about with respect to the query and under the overall context of a video. This mechanism is materialized into a family of general-purpose neural units and their multi-level architecture called Hierarchical Object-oriented Spatio-Temporal Reasoning (HOSTR) networks. This neural model maintains the objects' consistent lifelines in the form of a hierarchically nested spatio-temporal graph. Within this graph, the dynamic interactive object-oriented representations are built up along the video sequence, hierarchically abstracted in a bottom-up manner, and converge toward the key information for the correct answer. The method is evaluated on multiple major Video QA datasets and establishes new state-of-the-arts in these tasks. Analysis into the model's behavior indicates that object-oriented reasoning is a reliable, interpretable and efficient approach to Video QA., Comment: Accepted by IJCAI 2021. Please cite the conference version

In order to further enhance the computational efficiency and application scope of the bilinear functional links neural networks (BFLNN) filter, a pipelined BFLNN (PBFLNN) filter has been developed in this paper. The idea of the method is to divide the complex BFLNN structure into multiple simple BFLNN modules (with a smaller memory-length) and cascade connection in a pipelined fashion. Thanks to the simultaneous processing and the nested nonlinearity of the modules, the PBFLNN achieves a significant improvement in computation without degrading its performance. The simulation results have demonstrated the effectiveness of the proposed method and the potentials of the PBFLNN filter in many different applications.

Recently, there has been a growing interest in artificial neural networks, a rough simulation of the human brain's information processing ability, as modern and vastly sophisticated computational techniques. This interest has also been reflected in the water and wastewater field. This chapter presents a review of articles on the application of neural networks as useful tools to solve various problems in wastewater fields, especially those characterized by pollutants prediction. After a short description of theoretical background and practical basics concerning the computations performed employing neural networks, the essential water and wastewater applications of neural networks are demonstrated with suitable references. The considerable role of neural networks in prediction, simulation, optimization, and searching for the relationships between each parameter and treatment efficiency is discussed.

We consider the large sum of DC (Difference of Convex) functions minimization problem which appear in several different areas, especially in stochastic optimization and machine learning. Two DCA (DC Algorithm) based algorithms are proposed: stochastic DCA and inexact stochastic DCA. We prove that the convergence of both algorithms to a critical point is guaranteed with probability one. Furthermore, we develop our stochastic DCA for solving an important problem in multi-task learning, namely group variables selection in multi class logistic regression. The corresponding stochastic DCA is very inexpensive, all computations are explicit. Numerical experiments on several benchmark datasets and synthetic datasets illustrate the efficiency of our algorithms and their superiority over existing methods, with respect to classification accuracy, sparsity of solution as well as running time.

Over The Top technology (OTT) – thanks to the adaptive streaming, i.e. automatically adjusting the video stream’s quality – quickly takes over the role of its precedent IPTV and becomes the indispensable service for any provider of streaming video over the Internet. However in deployment, there are still several white spaces to improve the technology. In this paper we propose 3 techniques toward that goal. Firstly, taking into account the limited resource of the encoders, we propose a temporal and spatial determination of the pre-generated bitstreams based on the statistical analysis of network traffic. The number of streams remains unchanged but they fit better to the real condition of the transmission network. Secondly, addressing the booming number of available digital TV programs, we provide an adaptively customized catalogue of the TV programs. The broadcasted programs are organized by their contents’ natures. Viewers’ historical interaction with TV is incorporated to further fine-tuning the program lists proposed to viewers themselves. Finally, facing to the extreme situation when the OTT servers can still reach to the saturated state, we propose to switch adaptively between serve-client and peer-to-peer topologies to retrieve the requested TV program in an efficient way. These three adaptivity-methods are combined together in a full chain of OTT distribution system to be evaluated on their overall performance. The impact is expected to be pertinent.

This paper presents the design, implementation, and usage details of ARCADE, an extensible workbench for supporting the recovery of software systems' architectures, and for evaluating architectural change and decay. ARCADE has been developed and maintained over the past decade, and has been deployed in a number of research labs as well as within three large companies. ARCADE's implementation is available at https://bitbucket.org/joshuaga/arcade and the video depicting its use at https://tinyurl.com/arcade-tool-demo.

Video QA challenges modelers in multiple fronts. Modeling video necessitates building not only spatio-temporal models for the dynamic visual channel but also multimodal structures for associated information channels such as subtitles or audio. Video QA adds at least two more layers of complexity - selecting relevant content for each channel in the context of the linguistic query, and composing spatio-temporal concepts and relations in response to the query. To address these requirements, we start with two insights: (a) content selection and relation construction can be jointly encapsulated into a conditional computational structure, and (b) video-length structures can be composed hierarchically. For (a) this paper introduces a general-reusable neural unit dubbed Conditional Relation Network (CRN) taking as input a set of tensorial objects and translating into a new set of objects that encode relations of the inputs. The generic design of CRN helps ease the common complex model building process of Video QA by simple block stacking with flexibility in accommodating input modalities and conditioning features across both different domains. As a result, we realize insight (b) by introducing Hierarchical Conditional Relation Networks (HCRN) for Video QA. The HCRN primarily aims at exploiting intrinsic properties of the visual content of a video and its accompanying channels in terms of compositionality, hierarchy, and near and far-term relation. HCRN is then applied for Video QA in two forms, short-form where answers are reasoned solely from the visual content, and long-form where associated information, such as subtitles, presented. Our rigorous evaluations show consistent improvements over SOTAs on well-studied benchmarks including large-scale real-world datasets such as TGIF-QA and TVQA, demonstrating the strong capabilities of our CRN unit and the HCRN for complex domains such as Video QA., Major extension of our CVPR'20 paper to handle long video with text. arXiv admin note: substantial text overlap with arXiv:2002.10698

In this work, a specific performance measurement method known as Data Envelopment Analysis (DEA) is discussed for application in governmental management and improves the efficiency. DEA is a great model that is applied in many operating environments of the modern economy. According to Business Performance Improvement Resource (BPIR) definition, performance refers to current outputs and outcomes obtained from operating processes that permit evaluation and comparison of information. Performance can be expressed in non-financial and financial terms. Measurement refers to numerical information that quantifies input, output, and performance dimensions of processes, products, services, and the overall organization. In business, performance measurement helps managers improve their decision making and organization performance because it provides essential and quality feedback that the operations may be guided accordingly by allowing managers to achieve the best solution. It is a great way to understand, manage, and improve the overall functioning of the organization, especially to improve business success. If the measurement result is wrong or inaccurate, the users of data will be misled and can make a bad decision. First time introduced in 1970, DEA was quickly recognized as an excellent methodology for performance measurement. Being a “data oriented” approach for evaluating the performance of a set of peer entities called Decision Making Units (DMUs) which convert multiple inputs into multiple outputs, it is ideal for measuring the relative efficiencies of units with similar services or product and gives a big advantage of being able to deal with multidimensional nature of input/output variables.

In this paper, we present a solution to design a simulation model for mobile node with multiple radio interfaces in mobile ad-hoc wireless networks (MANET). This solution extends the mobile node architecture in NS2 network simulator to allow analysis and evaluation of the performance of multichannel mobile ad-hoc wireless networks. Simulation results show that the effectiveness of multi-radio mobile node model. Network performance with new mobile node architecture is greatly improved as the number of wireless interfaces increased.

We present Language-binding Object Graph Network, the first neural reasoning method with dynamic relational structures across both visual and textual domains with applications in visual question answering. Relaxing the common assumption made by current models that the object predicates pre-exist and stay static, passive to the reasoning process, we propose that these dynamic predicates expand across the domain borders to include pair-wise visual-linguistic object binding. In our method, these contextualized object links are actively found within each recurrent reasoning step without relying on external predicative priors. These dynamic structures reflect the conditional dual-domain object dependency given the evolving context of the reasoning through co-attention. Such discovered dynamic graphs facilitate multi-step knowledge combination and refinements that iteratively deduce the compact representation of the final answer. The effectiveness of this model is demonstrated on image question answering demonstrating favorable performance on major VQA datasets. Our method outperforms other methods in sophisticated question-answering tasks wherein multiple object relations are involved. The graph structure effectively assists the progress of training, and therefore the network learns efficiently compared to other reasoning models., Comment: Early version accepted by IJCAI20, Code available at https://github.com/thaolmk54/LOGNet-VQA