Search

Protein structure determination is a critical aspect of biological research, enabling us to understand protein function and potential applications. Recent advances in deep learning and artificial intelligence have led to the development of several protein structure prediction tools, such as AlphaFold2 and ColabFold. However, their performance has primarily been evaluated on well-characterised proteins, their performance on proteins lacking experimental structures, such as many snake venom toxins, has been less scrutinised. In this study, we evaluated three modelling tools on their prediction of over 1000 snake venom toxin structures for which no experimental structures exist. Our findings show that AlphaFold2 (AF2) performed the best across all assessed parameters. We also observed that ColabFold (CF) only scored slightly worse than AF2, while being computationally less intensive. All tools struggled with regions of intrinsic disorder, such as loops and propeptide regions, and performed well in predicting the structure of functional domains. Overall, our study highlights the importance of exercising caution when working with proteins with no experimental structures available, particularly those that are large and contain flexible regions. Nonetheless, leveraging computational structure prediction tools can provide valuable insights into the modelling of protein interactions with different targets and reveal potential binding sites, active sites, and conformational changes, as well as into the design of potential molecular binders for reagent, diagnostic, or therapeutic purposes.

Aim To determine neurodevelopmental outcome at 18 months after therapeutic hypothermia for hypoxic-ischaemic encephalopathy (HIE) infants in Vietnam, a low-middle-income country. Method Prospective cohort study investigating outcomes at 18 months in severely asphyxiated outborn infants who underwent therapeutic hypothermia for HIE in Hanoi, Vietnam, during the time period 2016-2019. Survivors were examined at discharge and at 6 and 18 months by a neonatologist, a neurologist and a rehabilitation physician, who were blinded to the infants' clinical severity during hospitalisation using two assessment tools: the Ages and Stages Questionnaire (ASQ) and the Hammersmith Infant Neurological Examination (HINE), to detect impairments and promote early interventions for those who require it. Results In total, 130 neonates, 85 (65%) with moderate and 45 (35%) with severe HIE, underwent therapeutic hypothermia treatment using phase change material. Forty-three infants (33%) died during hospitalisation and in infancy. Among the 87 survivors, 69 (79%) completed follow-up until 18 months. Nineteen children developed cerebral palsy (8 diplegia, 3 hemiplegia, 8 dyskinetic), and 11 had delayed neurodevelopment. At each time point, infants with a normal or delayed neurodevelopment had significantly higher ASQ and HINE scores (p<0.05) than those with cerebral palsy. Conclusion The rates of mortality and adverse neurodevelopment rate were high and comparable to recently published data from other low-middle-income settings. The ASQ and HINE were useful tools for screening and evaluation of neurodevelopment and neurological function.

Oligoclonal mixtures of broadly-neutralizing antibodies can neutralize complex compositions of similar and dissimilar antigens, making them versatile tools for the treatment of e.g., infectious diseases and animal envenomations. However, these biotherapeutics are complicated to develop due to their complex nature. In this work, we describe the application of various strategies for the discovery of cross-neutralizing nanobodies against key toxins in coral snake venoms using phage display technology. We prepare two oligoclonal mixtures of nanobodies and demonstrate their ability to neutralize the lethality induced by two North American coral snake venoms in mice, while individual nanobodies fail to do so. We thus show that an oligoclonal mixture of nanobodies can neutralize the lethality of venoms where the clinical syndrome is caused by more than one toxin family in a murine challenge model. The approaches described may find utility for the development of advanced biotherapeutics against snakebite envenomation and other pathologies where multi-epitope targeting is beneficial.

Via thousands of papers in Explainable AI (XAI), attention maps \cite{vaswani2017attention} and feature importance maps \cite{bansal2020sam} have been established as a common means for finding how important each input feature is to an AI's decisions. It is an interesting, unexplored question whether allowing users to edit the feature importance at test time would improve a human-AI team's accuracy on downstream tasks. In this paper, we address this question by leveraging CHM-Corr, a state-of-the-art, ante-hoc explainable classifier \cite{taesiri2022visual} that first predicts patch-wise correspondences between the input and training-set images, and then bases on them to make classification decisions. We build CHM-Corr++, an interactive interface for CHM-Corr, enabling users to edit the feature importance map provided by CHM-Corr and observe updated model decisions. Via CHM-Corr++, users can gain insights into if, when, and how the model changes its outputs, improving their understanding beyond static explanations. However, our study with 18 expert users who performed 1,400 decisions finds no statistical significance that our interactive approach improves user accuracy on CUB-200 bird image classification over static explanations. This challenges the hypothesis that interactivity can boost human-AI team accuracy and raises needs for future research. We open-source CHM-Corr++, an interactive tool for editing image classifier attention (see an interactive demo here: http://137.184.82.109:7080/). We release code and data on github: https://github.com/anguyen8/chm-corr-interactive., Comment: Accepted for presentation at the XAI4CV Workshop, part of the CVPR 2024 proceedings

Entrepreneurship is a global phenomenon because it contributes to economic growth, maintenance of social stability, and reduced unemployment. However, the role of entrepreneurial orientation in Vietnamese universities remains blurred. This study aims to understand the role of university education in shaping the entrepreneurial intention of Vietnamese students through entrepreneurial orientation, perceived desirability, and perceived feasibility. The PLS-SEM technique with SmartPLS 4 software tested the research model and hypotheses. The data set was collected from May 2023 to June 2023 via Google Forms with 411 respondents. The results show that entrepreneurial education has the most decisive impact on perceived desirability, followed by perceived feasibility, and both of these factors have a substantial impact on entrepreneurial orientation and entrepreneurial intention. Based on the results, the research has suggested practical implications that enhance entrepreneurial intention and promote the development of Vietnam's economy.

Cytotoxic T lymphocyte-associated antigen-4 (CTLA-4) is an immune checkpoint expressed in regulatory T (Treg) cells and activated T lymphocytes. Despite its potential as a treatment strategy for melanoma, CTLA-4 inhibition has limited efficacy. Using data from The Cancer Genome Atlas (TCGA) melanoma database and another dataset, we found that decreased CTLA4 mRNA was associated with a poorer prognosis in metastatic melanoma. To investigate further, we measured blood CTLA4 mRNA in 273 whole-blood samples from an Australian cohort and found that it was lower in metastatic melanoma than in healthy controls and associated with worse patient survival. We confirmed these findings using Cox proportional hazards model analysis and another cohort from the US. Fractionated blood analysis revealed that Treg cells were responsible for the downregulated CTLA4 in metastatic melanoma patients, which was confirmed by further analysis of published data showing downregulated CTLA-4 surface protein expression in Treg cells of metastatic melanoma compared to healthy donors. Mechanistically, we found that secretomes from human metastatic melanoma cells downregulate CTLA4 mRNA at the post-transcriptional level through miR-155 while upregulating FOXP3 expression in human Treg cells. Functionally, we demonstrated that CTLA4 expression inhibits the proliferation and suppressive function of human Treg cells. Finally, miR-155 was found to be upregulated in Treg cells from metastatic melanoma patients compared to healthy donors. Our study provides new insights into the underlying mechanisms of reduced CTLA4 expression observed in melanoma patients, demonstrating that post-transcriptional silencing of CTLA4 by miRNA-155 in Treg cells may play a critical role. Since CTLA-4 expression is downregulated in non-responder melanoma patients to anti-PD-1 immunotherapy, targeting miRNA-155 or other factors involved in regulating CTLA4 expression in Treg cells without affecting T cells could be a potenti

Image classifiers are information-discarding machines, by design. Yet, how these models discard information remains mysterious. We hypothesize that one way for image classifiers to reach high accuracy is to first zoom to the most discriminative region in the image and then extract features from there to predict image labels, discarding the rest of the image. Studying six popular networks ranging from AlexNet to CLIP, we find that proper framing of the input image can lead to the correct classification of 98.91% of ImageNet images. Furthermore, we uncover positional biases in various datasets, especially a strong center bias in two popular datasets: ImageNet-A and ObjectNet. Finally, leveraging our insights into the potential of zooming, we propose a test-time augmentation (TTA) technique that improves classification accuracy by forcing models to explicitly perform zoom-in operations before making predictions. Our method is more interpretable, accurate, and faster than MEMO, a state-of-the-art (SOTA) TTA method. We introduce ImageNet-Hard, a new benchmark that challenges SOTA classifiers including large vision-language models even when optimal zooming is allowed., Comment: NeurIPS 2023 Track on Datasets and Benchmarks

Many applications involve humans in the loop, where continuous and accurate human motion monitoring provides valuable information for safe and intuitive human-machine interaction. Portable devices such as inertial measurement units (IMUs) are applicable to monitor human motions, while in practice often limited computational power is available locally. The human motion in task space coordinates requires not only the human joint motion but also the nonlinear coordinate transformation depending on the parameters such as human limb length. In most applications, measuring these kinematics parameters for each individual requires undesirably high effort. Therefore, it is desirable to estimate both, the human motion and kinematic parameters from IMUs. In this work, we propose a novel computational framework for dual estimation in real-time exploiting in-network computational resources. We adopt the concept of field Kalman filtering, where the dual estimation problem is decomposed into a fast state estimation process and a computationally expensive parameter estimation process. In order to further accelerate the convergence, the parameter estimation is progressively computed on multiple networked computational nodes. The superiority of our proposed method is demonstrated by a simulation of a human arm, where the estimation accuracy is shown to converge faster than with conventional approaches.

The advent of deep learning has led to a significant gain in machine translation. However, most of the studies required a large parallel dataset which is scarce and expensive to construct and even unavailable for some languages. This paper presents a simple yet effective method to tackle this problem for low-resource languages by augmenting high-quality sentence pairs and training NMT models in a semi-supervised manner. Specifically, our approach combines the cross-entropy loss for supervised learning with KL Divergence for unsupervised fashion given pseudo and augmented target sentences derived from the model. We also introduce a SentenceBERT-based filter to enhance the quality of augmenting data by retaining semantically similar sentence pairs. Experimental results show that our approach significantly improves NMT baselines, especially on low-resource datasets with 0.46--2.03 BLEU scores. We also demonstrate that using unsupervised training for augmented data is more efficient than reusing the ground-truth target sentences for supervised learning., Comment: TMP and GN contributed equally

Robots performing human-scale manipulation tasks require an extensive amount of knowledge about their surroundings in order to perform their actions competently and human-like. In this work, we investigate the use of virtual reality technology as an implementation for robot environment modeling, and present a technique for translating scene graphs into knowledge bases. To this end, we take advantage of the Universal Scene Description (USD) format which is an emerging standard for the authoring, visualization and simulation of complex environments. We investigate the conversion of USD-based environment models into Knowledge Graph (KG) representations that facilitate semantic querying and integration with additional knowledge sources., Comment: 6 pages, 3 figures, ICRA 2024

Occlusion is a long-standing problem in computer vision, particularly in instance segmentation. ACM MMSports 2023 DeepSportRadar has introduced a dataset that focuses on segmenting human subjects within a basketball context and a specialized evaluation metric for occlusion scenarios. Given the modest size of the dataset and the highly deformable nature of the objects to be segmented, this challenge demands the application of robust data augmentation techniques and wisely-chosen deep learning architectures. Our work (ranked 1st in the competition) first proposes a novel data augmentation technique, capable of generating more training samples with wider distribution. Then, we adopt a new architecture - Hybrid Task Cascade (HTC) framework with CBNetV2 as backbone and MaskIoU head to improve segmentation performance. Furthermore, we employ a Stochastic Weight Averaging (SWA) training strategy to improve the model's generalization. As a result, we achieve a remarkable occlusion score (OM) of 0.533 on the challenge dataset, securing the top-1 position on the leaderboard. Source code is available at this https://github.com/nguyendinhson-kaist/MMSports23-Seg-AutoID.

Nearest neighbors (NN) are traditionally used to compute final decisions, e.g., in Support Vector Machines or k-NN classifiers, and to provide users with explanations for the model's decision. In this paper, we show a novel utility of nearest neighbors: To improve predictions of a frozen, pretrained classifier C. We leverage an image comparator S that (1) compares the input image with NN images from the top-K most probable classes; and (2) uses S's output scores to weight the confidence scores of C. Our method consistently improves fine-grained image classification accuracy on CUB-200, Cars-196, and Dogs-120. Also, a human study finds that showing lay users our probable-class nearest neighbors (PCNN) improves their decision accuracy over prior work which only shows only the top-1 class examples.

This paper provides an overview of the hardware and software components used in our test bed project the NET Playground. All source information is stored in the GitLab repository (https://gitlab.com/Paulteck/net-playground). In the Hardware section, we present sketches and 3D views of mechanical parts and technical drawings of printed boards. The Software section discusses relay control using shell scripts and the utilization of Ansible for automation. We also introduce a C++ framework for connecting with the INA231 energy sensor. This paper serves as a reference for understanding and replicating our project's hardware and software components.

Machine learning (ML) is increasingly being used in critical decision-making software, but incidents have raised questions about the fairness of ML predictions. To address this issue, new tools and methods are needed to mitigate bias in ML-based software. Previous studies have proposed bias mitigation algorithms that only work in specific situations and often result in a loss of accuracy. Our proposed solution is a novel approach that utilizes automated machine learning (AutoML) techniques to mitigate bias. Our approach includes two key innovations: a novel optimization function and a fairness-aware search space. By improving the default optimization function of AutoML and incorporating fairness objectives, we are able to mitigate bias with little to no loss of accuracy. Additionally, we propose a fairness-aware search space pruning method for AutoML to reduce computational cost and repair time. Our approach, built on the state-of-the-art Auto-Sklearn tool, is designed to reduce bias in real-world scenarios. In order to demonstrate the effectiveness of our approach, we evaluated our approach on four fairness problems and 16 different ML models, and our results show a significant improvement over the baseline and existing bias mitigation techniques. Our approach, Fair-AutoML, successfully repaired 60 out of 64 buggy cases, while existing bias mitigation techniques only repaired up to 44 out of 64 cases., Comment: In Proceedings of The 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE 2023)

The Margrabe Best-of-two (MBo2) strategy is a rule-based dynamic investment solution for the two-asset allocation problem. Its typical implementation involves yearly rebalancing the portfolio weights to 50–50 between a low-risk and high-risk asset. It uses intra-year weight adjustments to chase the momentum of the best performing asset by replicating the Margrabe formula for the value of a European option to exchange an asset for another asset at year-end. In practice, this means that the Margrabe portfolio allocation benefits from the upside potential of the high-risk asset and the downside protection from the low-risk asset. The MBo2 allocation depends on the assets’ prices, their return volatilities, and correlation, as well as the remaining time until year-end. In this paper, we derive analytical formulae and use simulations to provide insights into the sensitivity of the strategy's weights and performance to these input parameters. We also report the results of an extensive out-of-sample evaluation for the MBo2 strategy applied to the bond–equity, real estate–equity, and world equity–emerging market equity portfolio allocation problems.

In this article, we present the novel features of the recent version of SimAnMo, the Simulated Annealing Modeler. The tool creates models that correlate the size of one input parameter of an application to the corresponding runtime and thus SimAnMo allows predictions for larger input sizes. A focus lies on applications whose runtime grows exponentially in the input parameter size. Such programs are, for example, of high interest for cryptanalysis to analyze practical security of traditional and post‐quantum secure schemes. However, SimAnMo also generates reliable models for the widespread case of polynomial runtime behavior and also for the important case of factorial runtime increase. SimAnMo's model generation is based on a parallelized simulated annealing procedure and heuristically minimizes the costs of a model. Those may rely on different quality metrics. Insights into SimAnMo's software design and its usage are provided. We demonstrate the quality of SimAnMo's models for different algorithms from various application fields. We show that our approach also works well on ARM architectures.

Malware authors often use cryptographic tools such as XOR encryption and block ciphers like AES to obfuscate part of the malware to evade detection. Use of cryptography may give the impression that these obfuscation techniques have some provable guarantees of success. In this paper, we take a closer look at the use of cryptographic tools to obfuscate malware. We first find that most techniques are easy to defeat (in principle), since the decryption algorithm and the key is shipped within the program. In order to clearly define an obfuscation technique's potential to evade detection we propose a principled definition of malware obfuscation, and then categorize instances of malware obfuscation that use cryptographic tools into those which evade detection and those which are detectable. We find that schemes that are hard to de-obfuscate necessarily rely on a construct based on environmental keying. We also show that cryptographic notions of obfuscation, e.g., indistinghuishability and virtual black box obfuscation, may not guarantee evasion detection under our model. However, they can be used in conjunction with environmental keying to produce hard to de-obfuscate version of programs., Comment: This is the full version of the paper with the same title to appear in the Journal of Computer Virology and Hacking Techniques

Explaining the behaviors of deep neural networks, usually considered as black boxes, is critical especially when they are now being adopted over diverse aspects of human life. Taking the advantages of interpretable machine learning (interpretable ML), this work proposes a novel tool called Catastrophic Forgetting Dissector (or CFD) to explain catastrophic forgetting in continual learning settings. We also introduce a new method called Critical Freezing based on the observations of our tool. Experiments on ResNet articulate how catastrophic forgetting happens, particularly showing which components of this famous network are forgetting. Our new continual learning algorithm defeats various recent techniques by a significant margin, proving the capability of the investigation. Critical freezing not only attacks catastrophic forgetting but also exposes explainability., Comment: Master of Science Thesis at KAIST; 24 pages; Keywords: continual learning, catastrophic forgetting, XAI, attribution map, interpretability

Robust, reliable, and deterministic networks are essential for a variety of applications. In order to provide guaranteed communication network services, Time-Sensitive Networking (TSN) unites a set of standards for time-synchronization, flow control, enhanced reliability, and management. We design the TSN-FlexTest testbed with generic commodity hardware and open-source software components to enable flexible TSN measurements. We have conducted extensive measurements to validate the TSN-FlexTest testbed and to examine TSN characteristics. The measurements provide insights into the effects of TSN configurations, such as increasing the number of synchronization messages for the Precision Time Protocol, indicating that a measurement accuracy of 15 ns can be achieved. The TSN measurements included extensive evaluations of the Time-aware Shaper (TAS) for sets of Tactile Internet (TI) packet traffic streams. The measurements elucidate the effects of different scheduling and shaping approaches, while revealing the need for pervasive network control that synchronizes the sending nodes with the network switches. We present the first measurements of distributed TAS with synchronized senders on a commodity hardware testbed, demonstrating the same Quality-of-Service as with dedicated wires for high-priority TI streams despite a 200% over-saturation cross traffic load. The testbed is provided as an open-source project to facilitate future TSN research., Comment: 30 pages, 18 figures, 6 tables, IEEE TNSM, in print, 2024. Shorter version in print in IEEE Trans. on Network and Service Management (see related DOI below)

Explaining artificial intelligence (AI) predictions is increasingly important and even imperative in many high-stakes applications where humans are the ultimate decision-makers. In this work, we propose two novel architectures of self-interpretable image classifiers that first explain, and then predict (as opposed to post-hoc explanations) by harnessing the visual correspondences between a query image and exemplars. Our models consistently improve (by 1 to 4 points) on out-of-distribution (OOD) datasets while performing marginally worse (by 1 to 2 points) on in-distribution tests than ResNet-50 and a $k$-nearest neighbor classifier (kNN). Via a large-scale, human study on ImageNet and CUB, our correspondence-based explanations are found to be more useful to users than kNN explanations. Our explanations help users more accurately reject AI's wrong decisions than all other tested methods. Interestingly, for the first time, we show that it is possible to achieve complementary human-AI team accuracy (i.e., that is higher than either AI-alone or human-alone), in ImageNet and CUB image classification tasks., Comment: NeurIPS 2022 conference paper

Venomous snakes are important parts of the ecosystem, and their behavior and evolution have been shaped by their surrounding environments over the eons. This is reflected in their venoms, which are typically highly adapted for their biological niche, including their diet and defense mechanisms for deterring predators. Sub-Saharan Africa is rich in venomous snake species, of which many are dangerous to humans due to the high toxicity of their venoms and their ability to effectively deliver large amounts of venom into their victims via their bite. In this study, the venoms of 26 of sub-Saharan Africa's medically most relevant elapid and viper species were subjected to parallelized toxicovenomics analysis. The analysis included venom proteomics and in vitro functional characterization of whole venom toxicities, enabling a robust comparison of venom profiles between species. The data presented here corroborate previous studies and provide biochemical details for the clinical manifestations observed in envenomings by the 26 snake species. Moreover, two new venom proteomes (Naja anchietae and Echis leucogaster) are presented here for the first time. Combined, the presented data can help shine light on snake venom evolutionary trends and possibly be used to further improve or develop novel antivenoms.

Firewall policies are an important line of defence in cybersecurity, specifying which packets are allowed to pass through a network and which are not. These firewall policies are made up of a list of interacting rules. In practice, firewall can consist of hundreds or thousands of rules. This can be very difficult for a human to correctly configure. One proposed solution is to model firewall policies as Boolean expressions and use existing computer programs such as SAT solvers to verify that the firewall satisfies certain conditions. This paper takes an in-depth look at the Boolean expressions that represent firewall policies. We present an algorithm that translates a list of firewall rules into a Boolean expression in conjunctive normal form (CNF) or disjunctive normal form (DNF). We also place an upper bound on the size of the CNF and DNF that is polynomial in the number of rules in the firewall policy. This shows that past results suggesting a combinatorial explosion when converting from a Boolean expression in CNF to one in DNF does note occur in the context of firewall analysis

We construct a stochastic fluid process with an underlying piecewise deterministic Markov process (PDMP) akin to the one used in the construction of the rational arrival process (RAP) in Asmussen and Bladt (1999) which we call the RAP-modulated fluid process. As opposed to the classic Markov-modulated fluid process driven by a Markov jump process, the underlying PDMP of a RAP-modulated fluid process has a continuous state space and is driven by matrix parameters which may not be related to an intensity matrix. Through novel techniques we show how well-known formulae associated to the Markov-modulated fluid process, such as first passage probabilities and the stationary distribution of its queue, translate to its RAP-modulated counterpart.

This dataset contains various network monitoring channels of computing infrastructure collected during the period of almost 15 months in 2018/2019. It is aggregated over 3 different sliding windows.

Baffles and check-dam systems are often used as granular flow (rock avalanches, debris flows, etc.) control structures in regions prone to dangerous geological hazards leading to massive landslides. This paper explores the use of numerical modelling to simulate large volume granular flow and the effect of the presence of baffles and check dam systems on granular flow. In particular, the paper offers a solution based on the smoothed particle hydrodynamics numerical method, combined with a modified Bingham model with Mohr–Coulomb yield stress for granular flows. This method is parallelised at a large scale to perform high-resolution simulations of sand flowing down an inclined flume, obstructed by rigid control structures. We found that to maximise the flow deceleration ability of baffle arrays, the design of baffle height ought to reach a minimum critical value, which can be quantified from the flow depth without baffles (e.g. 2.7 times for frictional flows with friction angle of 27.5°). Also, the check-dam system was found to minimise run-out distances more effectively but experiences substantially higher forces compared to baffles. Finally, flow-control structures that resulted in lower run-out distances were associated with lower total energy dissipation, but faster kinetic energy dissipation in the granular flows; as well as lower downstream peak flow rates. © 2021, The Author(s), under exclusive licence to Springer-Verlag GmbH, DE part of Springer Nature.

Background and Purpose: The AFFINITY trial (Assessment of Fluoxetine in Stroke Recovery) reported that oral fluoxetine 20 mg daily for 6 months after acute stroke did not improve functional outcome and increased the risk of falls, bone fractures, and seizures. After trial medication was ceased at 6 months, survivors were followed to 12 months post-randomization. This preplanned secondary analysis aimed to determine any sustained or delayed effects of fluoxetine at 12 months post-randomization. Methods: AFFINITY was a randomized, parallel-group, double-blind, placebo-controlled trial in adults (n=1280) with a clinical diagnosis of stroke in the previous 2 to 15 days and persisting neurological deficit who were recruited at 43 hospital stroke units in Australia (n=29), New Zealand (4), and Vietnam (10) between 2013 and 2019. Participants were randomized to oral fluoxetine 20 mg once daily (n=642) or matching placebo (n=638) for 6 months and followed until 12 months after randomization. The primary outcome was function, measured by the modified Rankin Scale, at 6 months. Secondary outcomes for these analyses included measures of the modified Rankin Scale, mood, cognition, overall health status, fatigue, health-related quality of life, and safety at 12 months. Results: Adherence to trial medication was for a mean 167 (SD 48) days and similar between randomized groups. At 12 months, the distribution of modified Rankin Scale categories was similar in the fluoxetine and placebo groups (adjusted common odds ratio, 0.93 [95% CI, 0.76–1.14]; P =0.46). Compared with placebo, patients allocated fluoxetine had fewer recurrent ischemic strokes (14 [2.18%] versus 29 [4.55%]; P =0.02), and no longer had significantly more falls (27 [4.21%] versus 15 [2.35%]; P =0.08), bone fractures (23 [3.58%] versus 11 [1.72%]; P =0.05), or seizures (11 [1.71%] versus 8 [1.25%]; P =0.64) at 12 months. Conclusions: Fluoxetine 20 mg daily for 6 months after acute stroke had no delayed or sustained

This research addresses the question of which factors influencing the River Basin Organizations (RBOs) effectiveness in ensuring more cooperation and reducing conflict in the context of transboundary river basins. In order to answer this question, this research selected three river basins (Mekong, La Plata, and Okavango) to conduct research. The comparative approach of the qualitative methodology was applied to observe the variance between findings in three case studies. In each case study, the method of coding water events and gathering data/information from secondary sources are used to closely examine the linkage between RBOs institutional design, the extent of water events, RBOs effectiveness, and the contextual factors (political, socio-economic, and cultural). The findings have revealed that the organizational set-up of the RBOs has more influence in increasing RBOs effectiveness in the three case studies. Governance mechanisms have a higher impact on decreasing RBOs effectiveness compared to the organization set-up. In addition, the findings have shown that contextual factors have a strong interlinkage with both RBOs institutional design and RBOs effectiveness. The contextual factors can influence both terms in a direct/indirect and positive/negative way. Furthermore, RBOs institutional design, in turn, do matter in maximizing/minimizing the effects of the contextual factors on themselves. Ultimately, this research concluded that both organizational set-up and governance mechanisms of the RBOs in three case studies should be further enhanced in the future to strengthen the overall capacity of the system in dealing with the effects of contextual factors on conflict and cooperation in transboundary river basins.

In this paper, we demonstrate how to extend perfect spatial hashing (PSH) in order to hash multidimensional scientific data. As a use case we employ the problem domain of indexing nodes in a graph that represents Super Carbon Nanotubes (SCNTs). The goal of PSH is to hash multidimensional data without collisions. Since PSH results from the research on computer graphics, its principles and methods have only been tested on 2- and 3-dimensional problems. In our case, we need to hash up to 28 dimensions. In contrast to the original applications of PSH, we do not focus on GPUs as target hardware but on an efficient CPU implementation. Thus, this paper highlights the extensions to the original algorithm to make it suitable for higher dimensions. Comparing the compression and performance results of the new PSH based graphs and a structure-tailored custom data structure in our parallelized SCNT simulation software, we find that PSH in some cases achieves better compression by a factor of 1.7 while only increasing the total runtime by several percent. In particular, after our extension, PSH can also be employed to index sparse multidimensional scientific data from other domains where PSH can avoid additional index-structures like KD- or R-trees.

The aim of this study was to evaluate the effects of the substitution of maize silage (MS) for urea-treated rice straw (URS) in feedlot diets on finishing performance and body conformation traits of culled local Yellow cows; and to develop prediction equations for live weight (LW) of local Yellow cows using these body measurements. Twelve non-pregnant local Yellow cows (74.8 ± 23.6 months of age) were blocked by weight into 4 groups and then randomly allocated to one of three roughage treatments: 30% fresh maize + 70% URS (DM basis) (Control); 30% fresh maize + 40% URS + 30% MS (LMS); 30% fresh maize + 20% URS + 50% MS (HMS). Cows were individually supplemented with concentrate (1.1 kg DM/100 kg LW) and had free access to mixed roughage and water throughout the 12-week feeding trial, after a 2-week adaptation. Maize silage inclusion significantly increased daily feed intake and average daily gain (ADG) (P < 0.05). Moreover, the roughage containing 30% MS recorded the lowest feed conversion rate (7.8) and the highest final body condition score (3.43). However, LW and body measurements were not affected by the substitution (P > 0.05). Positive and significant correlations between LW and body measurements were observed. Regression analysis between chest girth (CG) and LW resulted in an equation with adjusted an R2 value of 82.0%. The multiple regression, which included CG, body length (BL) and withers height (WH) yielded the most reliable equation (adjusted R2 = 87.4%). The results indicate that MS can be a good alternative roughage source for feedlot culled local Yellow cows during winter in Northwest Vietnam. It is highly recommended that smallholder farmers raising non-pregnant Yellow cows use the regression equation: LW = 3.39 CG - 249 for prediction of LW in the absence of weighing scales.

Background: Emergence of cross-resistance to current anti-malarial drugs has led to an urgent need for identification of potential compounds with novel modes of action and anti-malarial activity against the resistant strains. One of the most promising therapeutic targets of anti-malarial agents related to food vacuole of malaria parasite is haemozoin, a product formed by the parasite through haemoglobin degradation. Methods: With this in mind, this study developed two-dimensional-quantitative structure–activity relationships (QSAR) models of a series of 21 haemozoin inhibitors to explore the useful physicochemical parameters of the active compounds for estimation of anti-malarial activities. The 2D-QSAR model with good statistical quality using partial least square method was generated after removing the outliers. Results: Five two-dimensional descriptors of the training set were selected: atom count (a_ICM); adjacency and distance matrix descriptor (GCUT_SLOGP_2: the third GCUT descriptor using atomic contribution to logP); average total charge sum (h_pavgQ) in pKa prediction (pH = 7); a very low negative partial charge, including aromatic carbons which have a heteroatom-substitution in “ortho” position (PEOE_VSA-0) and molecular descriptor (rsynth: estimating the synthesizability of molecules as the fraction of heavy atoms that can be traced back to starting material fragments resulting from retrosynthetic rules), respectively. The model suggests that the anti-malarial activity of haemozoin inhibitors increases with molecules that have higher average total charge sum in pKa prediction (pH = 7). QSAR model also highlights that the descriptor using atomic contribution to logP or the distance matrix descriptor (GCUT_SLOGP_2), and structural component of the molecules, including topological descriptors does make for better anti-malarial activity. Conclusions: The model is capable of predicting the anti-malarial activities of anti-haemozoin compounds. In addition, the, Malaria Journal, 20, art. no. 264; 2021

Today deep learning is widely used for building software. A software engineering problem with deep learning is that finding an appropriate convolutional neural network (CNN) model for the task can be a challenge for developers. Recent work on AutoML, more precisely neural architecture search (NAS), embodied by tools like Auto-Keras aims to solve this problem by essentially viewing it as a search problem where the starting point is a default CNN model, and mutation of this CNN model allows exploration of the space of CNN models to find a CNN model that will work best for the problem. These works have had significant success in producing high-accuracy CNN models. There are two problems, however. First, NAS can be very costly, often taking several hours to complete. Second, CNN models produced by NAS can be very complex that makes it harder to understand them and costlier to train them. We propose a novel approach for NAS, where instead of starting from a default CNN model, the initial model is selected from a repository of models extracted from GitHub. The intuition being that developers solving a similar problem may have developed a better starting point compared to the default model. We also analyze common layer patterns of CNN models in the wild to understand changes that the developers make to improve their models. Our approach uses commonly occurring changes as mutation operators in NAS. We have extended Auto-Keras to implement our approach. Our evaluation using 8 top voted problems from Kaggle for tasks including image classification and image regression shows that given the same search time, without loss of accuracy, Manas produces models with 42.9% to 99.6% fewer number of parameters than Auto-Keras' models. Benchmarked on GPU, Manas' models train 30.3% to 641.6% faster than Auto-Keras' models.

Asymptomatic surveillance testing together with COVID-19-related research can lead to positive SARS-CoV-2 tests resulting not from true infections, but non-infectious, non-hazardous by-products of research (amplicons). Amplicons can be widespread and persistent in lab environments and can be difficult to distinguish for true infections. We discuss prevention and mitigation strategies.

Modern power systems are experiencing the challenge of high uncertainty with the increasing penetration of renewable energy resources and the electrification of heating systems. In this paradigm shift, understanding electricity users' demand is of utmost value to retailers, aggregators, and policymakers. However, behind-the-meter (BTM) equipment and appliances at the household level are unknown to the other stakeholders mainly due to privacy concerns and tight regulations. In this paper, we seek to identify residential consumers based on their BTM equipment, mainly rooftop photovoltaic (PV) systems and electric heating, using imported/purchased energy data from utility meters. To solve this problem with an interpretable, fast, secure, and maintainable solution, we propose an integrated method called Interpretable Refined Motifs And binary Classification (IRMAC). The proposed method comprises a novel shape-based pattern extraction technique, called Refined Motif (RM) discovery, and a single-neuron classifier. The first part extracts a sub-pattern from the long time series considering the frequency of occurrences, average dissimilarity, and time dynamics while emphasising specific times with annotated distances. The second part identifies users' types with linear complexity while preserving the transparency of the algorithms. With the real data from Australia and Denmark, the proposed method is tested and verified in identifying PV owners and electrical heating system users., Comment: 22 pages, 13 figures

Desiccation cracking in clayey soils occurs when they lose moisture, leading to an increase in their compressibility and hydraulic conductivity and hence significant reduction of soil strength. The prediction of desiccation cracking in soils is challenging due to the lack of insights into the complex coupled hydro-mechanical process at the grain scale. In this paper, a new hybrid discrete-continuum numerical framework, capable of capturing hydro-mechanical behaviour of soil at both grain and macro scales, is proposed for predicting desiccation cracking in clayey soil. In this framework, a soil layer is represented by an assembly of DEM particles, each occupies an equivalent continuum space and carries physical properties governing unsaturated flow. These particles move freely in the computational space following the discrete element method (DEM), while their contact network and the continuum mixture theory are used to model the unsaturated flow. The dependence of particle-to-particle contact behaviour on water content is represented by a cohesive-frictional contact model, whose material properties are governed by the water content. In parallel with the theoretical development is a series of experiments on 3D soil desiccation cracking to determine essential properties and provide data for the validation of mechanical and physical behaviour. Very good agreement in both physical behaviour (e.g. evolution of water content) and mechanical behaviour (e.g. occurrence and development of cracks, and distribution of compressive and tensile strains) demonstrates that the proposed framework is capable of capturing the hydro-mechanical behaviour of soil during desiccation. The capability of the proposed framework facilitates numerical experiments for insights into the hydro-mechanical behaviour of unsaturated soils that have not been possible before.

Explaining the decisions of an Artificial Intelligence (AI) model is increasingly critical in many real-world, high-stake applications. Hundreds of papers have either proposed new feature attribution methods, discussed or harnessed these tools in their work. However, despite humans being the target end-users, most attribution methods were only evaluated on proxy automatic-evaluation metrics (Zhang et al. 2018; Zhou et al. 2016; Petsiuk et al. 2018). In this paper, we conduct the first user study to measure attribution map effectiveness in assisting humans in ImageNet classification and Stanford Dogs fine-grained classification, and when an image is natural or adversarial (i.e., contains adversarial perturbations). Overall, feature attribution is surprisingly not more effective than showing humans nearest training-set examples. On a harder task of fine-grained dog categorization, presenting attribution maps to humans does not help, but instead hurts the performance of human-AI teams compared to AI alone. Importantly, we found automatic attribution-map evaluation measures to correlate poorly with the actual human-AI team performance. Our findings encourage the community to rigorously test their methods on the downstream human-in-the-loop applications and to rethink the existing evaluation metrics., Comment: NeurIPS 2021; 10 pages of Main text; 28 pages of Appendix

Ho Chi Minh  University of Food Industry (HUFI) is gradually reforming and updating English teaching methods to enhance the learners’ quality. Within this context, foreign teachers are required to teach in English speaking classes with the assistance and observation of native teachers. This study developed factors that foreign teacher influencing the learner’s favour with English. An online google form questionnaire was conducted on participants called students at HUFI. The respondents answer an open-ended questionnaire. The results will benefit English-speaking-foreign teachers (ESFTs) in terms of realizing their strengths and weaknesses. This study aimed to develop solutions to minimize disadvantageous factors and maximize benefits to motivate students toward learning English in classes with foreign teachers.&nbsp

[EN] Massive content delivery is in the spotlight of the research community as both data traffic and the number of connected mobile devices are increasing at an incredibly fast pace. The enhanced mobile broadband (eMBB) is one of the main use cases for the fifth generation of mobile networks (5G), which focuses on transmitting greater amounts of data at higher data rates than in the previous generations, but also on increasing the area capacity (given in bits per second per square meter) and reliability. However, the broadcast and multicast implementation in 5G and presents several drawbacks such as unexpected disconnections and the lack of device-specific QoS guarantees. As a result, whenever the exact same content is to be delivered to numerous mobile devices simultaneously, this content must be replicated. Hence, the same number of parallel unicast sessions as users are needed. Therefore, novel systems that provide efficient massive content delivery and reduced energy consumption are needed. In this paper, we present a network-coded cooperation (NCC) protocol for efficient massive content delivery and the analytical model that describes its behavior. The NCC protocol combines the benefits of cooperative architectures known as mobile clouds (MCs) with Random Linear Network Coding (RLNC). Our results show the benefits of our NCC protocol when compared to the establishment of numerous parallel unicast sessions are threefold: offload data traffic from the cellular link, reduce the energy consumption at the cooperating users, and provide throughput gains when the cellular bandwidth is insufficient.

In this paper we propose a distributed architecture to provide machine learning practitioners with a set of tools and cloud services that cover the whole machine learning development cycle: ranging from the models creation, training, validation and testing to the models serving as a service, sharing and publication. In such respect, the DEEP-Hybrid-DataCloud framework allows transparent access to existing e-Infrastructures, effectively exploiting distributed resources for the most compute-intensive tasks coming from the machine learning development cycle. Moreover, it provides scientists with a set of Cloud-oriented services to make their models publicly available, by adopting a serverless architecture and a DevOps approach, allowing an easy share, publish and deploy of the developed models.

The research aims to identify the components of brand equity that affect consumer purchasing intentions and measure the effect of brand equity components on the intention of consumer purchases at the convenience stores in Ho Chi Minh City. The authors conduct the group discussions, expert discussion, and then analyze data from 200 valid questionnaires with four components of brand equity, namely perceived quality, brand loyalty, brand association, brand awareness. The results of Exploratory Factor Analysis (EFA) show that all four elements have a positive effect on the customer' intention to purchase in the convenience store of Ho Chi Minh City. In particular, the brand association factor has the strongest influence, followed by perceived quality, brand awareness, and brand loyalty. This research contributes that the results confirm the theory of Aaker (1991), Brown and Stayman (1992), Cobb-Walgren et al. (1995), MacKenzie (1986) in the new context of convenience stores in Ho Chi Minh City, Vietnam. Besides, the study gives some recommendations to help convenience stores improve the elements of brand equity and it, to enhance attraction for consumers.

Key Points: The achievement of SDG 3 depends on many other SDGs; some SDGs are logically inconsistent, especially in the attempt to increase conventionally defined GDP while preserving natural capital. Any short-term gains for human health from further forest conversion (e.g. food production) creates short- and long-term, direct and indirect health risks for humans, as well as for other biota. Failure to ensure universal access to sexual and reproductive healthcare services (including family planning) will increase pressure on forests at local, regional and global scales. The burning and clearing of forests cause significant harm to health via impaired quality of water, soil and air; increased exposure to infectious diseases and impacts climate regulation. Many infectious diseases are associated with forest disturbances and intrusions; some important infectious diseases have emerged from forests (notably HIV/AIDS). Greater exposure to green space, including forests, provides mental and physical health benefits for the growing global urban population.

The changes in coloration with age in males and females of Coeliccia mattii Phan & Kompier, 2016 from the Central Highlands of Vietnam are documented. Immature forms have extensive yellow markings on the thorax and S9 is yellowish, while in the fully mature male form, the thorax is largely black and covered in pruinosity and S9 is completely black. S9 is yellowish in the immature female while S9 is black in the fully mature female.