Search

Emotion analysis of students plays an important role in teaching effect evaluation. To develop robust algorithms for emotion analysis of students, a database from real classrooms is required. However, most existing databases were collected from adults and constructed in laboratory settings. In this paper, we present a manually-annotated facial action unit database from juveniles in real classrooms. Our database has three main characteristics: (1) it provides numerous education-related action units data from primary and high schools, complementing the vacancy of the publicly available educational action unit databases; (2) it contains 256,220 manually-annotated facial images of 1,796 juveniles, frame-by-frame annotated with 12 action units and 6-level intensities for each action unit; (3) it covers many challenges in the wild, including different head poses, facial resolution, illumination, and occlusion, supplementing action unit databases in the wild for research. The baseline for action unit detection and action unit intensity estimation are provided for future references. Especially, we apply weighted balance loss to solve imbalance within and between labels. Our database will be available to the research community: http://www.dlc.sjtu.edu.cn/rfau database.

Protein docking protocols typically involve global docking scan, followed by re-ranking of the scan predictions by more accurate scoring functions that are either computationally too expensive or algorithmically impossible to include in the global scan. Development and validation of scoring methodologies are often performed on scoring benchmark sets (docking decoys) which offer concise and nonredundant representation of the global docking scan output for a large and diverse set of protein-protein complexes. Two such protein-protein scoring benchmarks were built for the Dockground resource, which contains various datasets for the development and testing of protein docking methodologies. One set was generated based on the Dockground unbound docking benchmark 4, and the other based on protein models from the Dockground model-model benchmark 2. The docking decoys were designed to reflect the reality of the real-case docking applications (e.g., correct docking predictions defined as near-native rather than native structures), and to minimize applicability of approaches not directly related to the development of scoring functions (reducing clustering of predictions in the binding funnel and disparity in structural quality of the near-native and non-native matches). The sets were further characterized by the source organism and the function of the protein-protein complexes. The sets, freely available to the research community on the Dockground webpage, present a unique, user-friendly resource for the developing and testing of protein-protein scoring approaches.

BackgroundProstate Cancer is the second lethal malignancy in men worldwide. In the past, numerous research groups investigated the omics profiles of patients and scrutinized biomarkers for the diagnosis and prognosis of prostate cancer. However, information related to the biomarkers is widely scattered across numerous resources in complex textual format, which poses hindrance to understand the tumorigenesis of this malignancy and scrutinization of robust signature. To the best of authors’ knowledge, there is no resource that can consolidate the information contained in all the published literature.ResultsHere, we present ProCanBio, a manually curated database that maintains detailed data on 2053 entries of potential prostate cancer biomarkers obtained from 412 publications in user friendly tabular format. Among them, 766 protein-based, 488 RNA-based, 157 genomic mutations, 261 miRNA-based, and 122 are metabolites-based biomarkers. To explore the information in the resource, a web-based interactive platform was developed with searching, and browsing facilities. ProCanBio is freely available and is compatible with most web browsers and devices. Eventually, we anticipated this resource will be highly useful for the research community involved in the area of prostate malignancy.AvailabilityProCanBio is available at: https://webs.iiitd.edu.in/raghava/procanbio/

Cross-Lingual Text Reuse Detection (CLTRD) has recently attracted the attention of the research community due to a large amount of digital text readily available for reuse in multiple languages through online digital repositories. In addition, efficient machine translation systems are freely and readily available to translate text from one language into another, which makes it quite easy to reuse text across languages, and consequently difficult to detect it. In the literature, the most prominent and widely used approach for CLTRD is Translation plus Monolingual Analysis (T+MA). To detect CLTR for English-Urdu language pair, T+MA has been used with lexical approaches, namely, N-gram Overlap, Longest Common Subsequence, and Greedy String Tiling. This clearly shows that T+MA has not been thoroughly explored for the English-Urdu language pair. To fulfill this gap, this study presents an in-depth and detailed comparison of 26 approaches that are based on T+MA. These approaches include semantic similarity approaches (semantic tagger based approaches, WordNet-based approaches), probabilistic approach (Kullback-Leibler distance approach), monolingual word embedding-based approaches siamese recurrent architecture, and monolingual sentence transformer-based approaches for English-Urdu language pair. The evaluation was carried out using the CLEU benchmark corpus, both for the binary and the ternary classification tasks. Our extensive experimentation shows that our proposed approach that is a combination of 26 approaches obtained an F 1 score of 0.77 and 0.61 for the binary and ternary classification tasks, respectively, and outperformed the previously reported approaches [ 41 ] ( F 1 = 0.73) for the binary and ( F 1 = 0.55) for the ternary classification tasks) on the CLEU corpus.

Authentication receives enormous consideration from the research community and is proven to be an interesting field in today’s era. User authentication is the major concern because people have their private data on devices. To strengthen user authentication, passwords have been introduced. In the past, the text-based password was the traditional way of authentication, but this method has particular shortcomings. The graphical password has been introduced as an alternative, which uses a picture or a set of pictures to generate a password. In the future, it is a requirement of such approaches to maintain robustness and consume fewer energy resources to become suitable for the Green Internet of Things (IoT). Similarly, diverse graphical password authentication mechanisms have been used to provide users with better security and usability. In this article, we conduct an extensive survey on the existing approaches of graphical password authentication to highlight the challenges required to be addressed for Green IoT. In comparison to other existing surveys, the objective is to consolidate the graphical password technique and to identify the problem associated with it. Besides, this survey will also identify the vulnerabilities of the graphical password against several potential attacks. We have also examined the strengths and weaknesses of each technique along with the future research directions. This study also evaluates the usability of each approach by considering learnability, memorability, and so forth and also presents a comparative analysis with security.

Background & Objective: Ayurveda, the “Mother of all healing”, has existed for over 5,000 years and hence is considered to be the oldest healing science. Ayurveda states that the mind can heal and transform a person's whole being as the mind and body are associated. Herbs are the heart of Ayurvedic belief. They are used to boost defense against diseases and viruses and keep the brain, body, and soul in complete balance. Although ayurvedic medicines and herbs have natural components, they should still be used with certain precautions under the supervision of a medical practitioner. This study aims to manually curate information for the various ayurvedic medicinal herbs that have antiviral activity against harmful viruses. Methods: Detailed information is collected from the literature regarding the following (a) types of viruses (b) which particular category they belong to(c) the respective components of herbs that are responsible for curing viruses. We developed a web interface with the help of php and mysql to get the desired output. Results: The database consists of 104 viruses and 704 natural components. The web server is available at: http://ayurvir.com. Interpretation & Conclusion: We believe that AyurVirDB database will be extremely beneficial for the research community. It not only aids in investigations of Ayurvedic medicinal plants and their components. On the emergence or re-emergence of a virus, one could be able to predict the ayurvedic plants/herbs used for viral treatment based on virus similarity or disease symptoms.

As a widely known disease diabetes mellitus makes the human body produce quite less hormone and also tend to cause increased glucose that results in abnormal metabolism of varied organs in the body like eyes, kidneys, etc. Diabetic analysis has attracted the research community to treat some missing values and class imbalance issues. The performance of diabetes mellitus classification by the usage of machine learning techniques is comparatively low. We suggest this paper on imbalanced dataset with missing values, an efficient prediction algorithm for diabetes mellitus classification using Deep 1D-Convolutional Neural Network values. The outlier detection is used for removing missing values first. Then, oversampling method (SMOTE) is used to reduce the influence of imbalance class on prediction performance. Finally, predictions are produced using a DCNN classifier and are evaluated using a selective set of evaluation indicators. Experiments on the Pima Indian diabetes dataset (PIDD) from UCI Repository (University of California at Irvine) have yielded positive results. Our proposed DCNN algorithm has been shown to be successful and superior.

Robotic manipulation is fundamental to many realworld applications; however, it is an unsolved problem that remains a very active research area. New algorithms for robot perception and control are frequently proposed by the research community. These methods must be thoroughly evaluated in realistic conditions before they can be adopted by industry. This process can be extremely time consuming, mainly due to the complexity of integrating different hardware and software components.

Food has a direct impact on an individual's life and is a significant area for the research community. Computational techniques in food-related computing are performed to address various food-related issues in the field of agronomy, medicine, biology etc. In this paper, we present a comprehensive review of research exclusively on segmentation techniques used for food computing. This paper illustrates the viable segmentation techniques used for food image segmentation. It also provides a comprehensive review of the same. A relevant survey on 66 research papers has been done to provide different food image segmentation techniques. A comparative study among these techniques is also done based on different parameters like type of algorithm, segmentation technique, dataset, and accuracy. Moreover, this paper focuses on research challenges in food recognition. Also a framework has been proposed in this paper to overcome the problem of watershed and OTSU algorithm.

Facial feature extraction (FFE) is considered as a challenging area for computer vision and artificial intelligence research community. There are numerous application domains of face recognition such as demographic classification and facial disease classification. In last few years, numerous feature extraction approaches are proposed. This review focuses on studies that exclusively use facial asymmetry as one of the main subject-specific facial characteristics for FFE. This paper provides a review about the related research conducted in the past two decades. First, we summarize the conventional FFE approaches and their main algorithms. Application of deep networks to facial asymmetry based FFE approaches is then presented. Multi-network deep models suitable for asymmetry-based FFE for different applications are also focused in this review. We presented the details about the publicly available face datasets, evaluation metrics, and comparison of the state-of-the-art results is also presented. The directions of asymmetry-based FFE for future research is also presented to provide an awareness about the existing and future trends. This review is presented to provide directions and ideas for future research in the field of facial asymmetry.

A novel radiological research field pursuing comprehensive quantitative image, namely “Radiomics,” gained traction along with the advancement of computational technology and artificial intelligence. This novel concept for analyzing medical images brought extensive interest to the neuro-oncology and neuroradiology research community to build a diagnostic workflow to detect clinically relevant genetic alteration of gliomas noninvasively. Although quite a few promising results were published regarding MRI-based diagnosis of isocitrate dehydrogenase (IDH) mutation in gliomas, it has become clear that an ample amount of effort is still needed to render this technology clinically applicable. At the same time, many significant insights were discovered through this research project, some of which could be “reverse engineered” to improve conventional non-radiomic MR image acquisition. In this review article, the authors aim to discuss the recent advancements and encountering issues of radiomics, how we can apply the knowledge provided by radiomics to standard clinical images, and further expected technological advances in the realm of radiomics and glioma.

Accessible chromatin is a highly informative structural feature for identifying regulatory elements, which provides a large amount of information about transcriptional activity and gene regulatory mechanisms. Human ATAC-seq datasets are accumulating rapidly, prompting an urgent need to comprehensively collect and effectively process these data. We developed a comprehensive human chromatin accessibility database (ATACdb, http://www.licpathway.net/ATACdb), with the aim of providing a large amount of publicly available resources on human chromatin accessibility data, and to annotate and illustrate potential roles in a tissue/cell type-specific manner. The current version of ATACdb documented a total of 52 078 883 regions from over 1400 ATAC-seq samples. These samples have been manually curated from over 2200 chromatin accessibility samples from NCBI GEO/SRA. To make these datasets more accessible to the research community, ATACdb provides a quality assurance process including four quality control (QC) metrics. ATACdb provides detailed (epi)genetic annotations in chromatin accessibility regions, including super-enhancers, typical enhancers, transcription factors (TFs), common single-nucleotide polymorphisms (SNPs), risk SNPs, eQTLs, LD SNPs, methylations, chromatin interactions and TADs. Especially, ATACdb provides accurate inference of TF footprints within chromatin accessibility regions. ATACdb is a powerful platform that provides the most comprehensive accessible chromatin data, QC, TF footprint and various other annotations.

The service of authentication constitutes the spine of all security properties. It is the phase where entities prove their identities to each other and generally establish and derive cryptographic keys to provide confidentiality, data integrity, non-repudiation, and availability. Due to the heterogeneity and the particular security requirements of IoT (Internet of Things), developing secure, low-cost, and lightweight authentication protocols has become a serious challenge. This has excited the research community to design and develop new authentication protocols that meet IoT requirements. An interesting hardware technology, called PUFs (Physical Unclonable Functions), has been the subject of many subsequent publications on lightweight, low-cost, and secure-by-design authentication protocols. This has turned our attention to investigate the most recent PUF-based authentication protocols for IoT. In this paper, we review the security of these protocols. We first provide the necessary background on PUFs, their types, and related attacks. Also, we discuss how PUFs are used for authentication. Then, we analyze the security of PUF-based authentication protocols to identify and report common security issues and design flaws, as well as to provide recommendations for future authentication protocol designers.

Growing awareness about sustainable development and the environmental problems involved in using nonbiodegradable materials has motivated the research community to develop environment-friendly materials. Developments have been achieved in the field of natural fibers and biopolymers, still there remain unanswered questions regarding the high-quality cost-effective manufacturing of natural fiber reinforced composites. The natural fiber-based polymeric composites are being used extensively in engineering applications, especially in the nonstructural parts and components. Near-net processing techniques such as compression molding, extrusion, and injection molding are well-developed for natural fiber reinforced composites. However, secondary processes such as joining, machining, and surface modification are still unexplored and need to be investigated in detail. The present research endeavor is an attempt to experimentally investigate the adhesive joining behavior of jute/sisal reinforced epoxy composites. The laminates based on three different material configurations in woven mat form, namely, pure jute, pure sisal, and hybrid jute/sisal reinforced epoxy have been fabricated by hand layup process. Different lap joint configurations with through holes in adherends overlapping area have been investigated. It has been established experimentally that the holes in the adherends provide a hinge-effect in the overlapping area and help in defining the failure load of the composite joint. The different arrangement of holes has been investigated and the best design of hole arrangement has been proposed for adhesive joining of jute/sisal fiber reinforced epoxy laminates. It was found that the holes (filled with an epoxy adhesive) in the overlap area result in 6–18% improvement in the failure load for different materials as compared to the joints with only adhesive bonding. Moreover, the field-emission scanning electron microscopy micrographs have been used to understand the failure mechanism of the adhesively bonded natural fiber reinforced composite laminates.

Due to the complex and dynamic environment of social media, user generated contents (UGCs) may inadvertently leak users’ personal aspects, such as the personal attributes, relationships and even the health condition, and thus place users at high privacy risks. Limited research efforts, thus far, have been dedicated to the privacy detection from users’ unstructured data (i.e., UGCs). Moreover, existing efforts mainly focus on applying conventional machine learning techniques directly to traditional hand-crafted privacy-oriented features, ignoring the powerful representing capability of the advanced neural networks. In light of this, in this article, we present a fine-grained privacy detection network (GrHA) equipped with graph-regularized hierarchical attentive representation learning. In particular, the proposed GrHA explores the semantic correlations among personal aspects with graph convolutional networks to enhance the regularization for the UGC representation learning, and, hence, fulfil effective fine-grained privacy detection. Extensive experiments on a real-world dataset demonstrate the superiority of the proposed model over state-of-the-art competitors in terms of eight standard metrics. As a byproduct, we have released the codes and involved parameters to facilitate the research community.

Demand is present among security practitioners for improving cyber situational awareness (SA), but capability and assessment have not risen to match. SA is an integral component of cybersecurity for everyone from individuals to business to response teams and threat exchanges. In this Field Note, we highlight existing research and our field observations, a recent review of cyber SA research literature, and call upon the research community to help address three research problems in situational awareness for cybersecurity. The gaps suggest the need to (1) understand what cyber SA is from the human operators’ perspectives, then (2) measure it so that (3) the community can learn whether SA makes a difference in meaningful ways to cybersecurity, and whether methods, technology, or other solutions would improve SA and thus, improve those outcomes.

Certain research strands can yield “forbidden knowledge”. This term refers to knowledge that is considered too sensitive, dangerous or taboo to be produced or shared. Discourses about such publication restrictions are already entrenched in scientific fields like IT security, synthetic biology or nuclear physics research. This paper makes the case for transferring this discourse to machine learning research. Some machine learning applications can very easily be misused and unfold harmful consequences, for instance, with regard to generative video or text synthesis, personality analysis, behavior manipulation, software vulnerability detection and the like. Up till now, the machine learning research community embraces the idea of open access. However, this is opposed to precautionary efforts to prevent the malicious use of machine learning applications. Information about or from such applications may, if improperly disclosed, cause harm to people, organizations or whole societies. Hence, the goal of this work is to outline deliberations on how to deal with questions concerning the dissemination of such information. It proposes a tentative ethical framework for the machine learning community on how to deal with forbidden knowledge and dual-use applications.

As the adoption rate of commercial smart home solutions increases, it drives the development of novel system features, needed to support advanced user scenarios. Being able to remotely control the household is not enough already, and the efforts are made to provide context-aware and intelligent homes, which detect user activities, learn about user habits, adapt to particular users, create intelligent alarms, seamlessly integrate with remote services, etc. On the other hand, significant efforts have already been made by the research community to identify data mining scenarios applicable to smart home solutions, and to propose and improve the algorithms suitable for this purpose. This article summarizes state-of-the-art research in the field of machine learning for smart home solutions.

Over the past decade, the available crystal structures have almost doubled in Protein Data Bank (PDB) providing the research community with a series of similar crystal structures to choose from for future docking studies. With the steady growth in the number of high-resolution three-dimensional protein structures, ligand docking-based virtual screening of chemical libraries to a receptor plays a critical role in the drug discovery process by identifying new drug candidates. Thus, identifying potential candidates among all the available structures in a database for docking studies is of utmost importance. Our work examined whether one could use the resolution of a number of known structures, without considering other parameters, to choose a good experimental structure for various docking studies to find more useful drug leads. We expected that a good experimental structure for docking studies to be the one that gave favorable docking with the largest number of ligands among the experimental structures to be selected. We chose three protein test systems for our study, all belonging to the family of MAPK: (1) JNK1, (2) JNK2, and (3) JNK3. On analysis of the results, the best resolution structures showed significant variations from the expected values in their result, whereas the poor resolution structures proved to be better candidates for docking studies.

The problem of network intrusion detection poses innumerable challenges to the research community, industry, and commercial sectors. Moreover, the persistent attacks occurring on the cyber-threat landscape compel researchers to devise robust approaches in order to address the recurring problem. Given the presence of massive network traffic, conventional machine learning algorithms when applied in the field of network intrusion detection are quite ineffective. Instead, a hybrid multimodel solution when sought improves performance thereby producing reliable predictions. Therefore, this article presents an ensemble model using metaclassification approach enabled by stacked generalization. Two contemporary as well as heterogeneous datasets, namely, UNSW NB-15, a packet-based dataset, and UGR’16, a flow-based dataset, that were captured in emulated as well as real network traffic environment, respectively, were used for experimentation. Empirical results indicate that the proposed stacking ensemble is capable of generating superior predictions with respect to a real-time dataset (97% accuracy) than an emulated one (94% accuracy).

Background A standard or consensus definition of a systematic review does not exist. Therefore, if there is no definition about a systematic review in secondary studies that analyse them or the definition is too broad, inappropriate studies might be included in such evidence synthesis. The aim of this study was to analyse the definition of a systematic review (SR) in health care literature, elements of the definitions that are used and to propose a starting point for an explicit and non-ambiguous SR definition. Methods We included overviews of systematic reviews (OSRs), meta-epidemiological studies and epidemiology textbooks. We extracted the definitions of SRs, as well as the inclusion and exclusion criteria that could indicate which definition of a SR the authors used. We extracted individual elements of SR definitions, categorised and quantified them. Results Among the 535 analysed sources of information, 188 (35%) provided a definition of a SR. The most commonly used reference points for the definitions of SRs were Cochrane and the PRISMA statement. We found 188 different elements of SR definitions and divided them into 14 categories. The highest number of SR definition elements was found in categories related to searching (N = 51), analysis/synthesis (N = 23), overall methods (N = 22), quality/bias/appraisal/validity (N = 22) and aim/question (N = 13). The same five categories were also the most commonly used combination of categories in the SR definitions. Conclusion Currently used definitions of SRs are vague and ambiguous, often using terms such as clear, explicit and systematic, without further elaboration. In this manuscript we propose a more specific definition of a systematic review, with the ultimate aim of motivating the research community to establish a clear and unambiguous definition of this type of research.

It's widely appreciated that liquid–liquid phase separation (LLPS) underlies the formation of membraneless organelles, which function to concentrate proteins and nucleic acids. In the past few decades, major efforts have been devoted to identify the phase separation associated proteins and elucidate their functions. To better utilize the knowledge dispersed in published literature, we developed PhaSepDB (http://db.phasep.pro/), a manually curated database of phase separation associated proteins. Currently, PhaSepDB includes 2914 non-redundant proteins localized in different organelles curated from published literature and database. PhaSepDB provides protein summary, publication reference and sequence features of phase separation associated proteins. The sequence features which reflect the LLPS behavior are also available for other human protein candidates. The online database provides a convenient interface for the research community to easily browse, search and download phase separation associated proteins. As a centralized resource, we believe PhaSepDB will facilitate the future study of phase separation.

Keeping firearms inside an arsenal with adequate safety is one of the prime concerns of the research community considering limitless damage in case of falling the firearms to the stain accused or rogue peoples in the society. In this regard, a remotely-operated, low-cost, and high-accuracy solution for maintaining safety in storing the firearms inside an arsenal is of utmost importance specially from the perspective of developing countries. Such a solution is yet to be proposed in the literature to the best of our knowledge. Therefore, in this study, we propose a new solution that can provide safety of firearms inside an arsenal in real time using low-power COTS sensor-based safety mechanism. Here, we propose judicious placing of RFID tags and RFID antennas, and develop a software module to retain the hardware related components inter-operable with each other. Besides, we develop a lock-and-key technique that exhibits a high standard of safety for the firearms inside the arsenal. Our solution also presents a time-attendance management system. Hardware part of our real implementation comprises a motor unit and a RFID reader, which is a low-frequency reader and connected to a host computer machine. All these components combined together provide several functionalities such as displaying ID, registering name, time of drawing firearms from the arsenal and so on. We present the implemented solution to real professional users. User evaluation of the proposed system confirms efficacy of our solution in real usage.

More accurate and more complete predictions of cis-regulatory modules (CRMs) and constituent transcriptional factor (TF) binding sites (TFBSs) in genomes can facilitate characterizing functions of regulatory sequences. Here, we developed a database PCRMS (https://cci-bioinfo.uncc.edu) that stores highly accurate and unprecedentedly complete maps of predicted CRMs and TFBSs in the human and mouse genomes. The web interface allows the user to browse CRMs and TFBSs in an organism, find the closest CRMs to a gene, search CRMs around a gene, and find all TFBSs of a TF. PCRMS can be a useful resource for the research community to characterize regulatory genomes.

The outbreak of the novel coronavirus disease 2019 (COVID-19), caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has been continuously affecting human lives and communities around the world in many ways, from cities under lockdown to new social experiences. Although in most cases COVID-19 results in mild illness, it has drawn global attention due to the extremely contagious nature of SARS-CoV-2. Governments and healthcare professionals, along with people and society as a whole, have taken any measures to break the chain of transition and flatten the epidemic curve. In this study, we used multiple data sources, i.e., PubMed and ArXiv, and built several machine learning models to characterize the landscape of current COVID-19 research by identifying the latent topics and analyzing the temporal evolution of the extracted research themes, publications similarity, and sentiments, within the time-frame of January- May 2020. Our findings confirm the types of research available in PubMed and ArXiv differ significantly, with the former exhibiting greater diversity in terms of COVID-19 related issues and the latter focusing more on intelligent systems/tools to predict/diagnose COVID-19. The special attention of the research community to the high-risk groups and people with complications was also confirmed.

Traffic forecasting is an important and challenging problem for intelligent transportation systems due to the complex spatial dependencies among neighboring roads and changing road conditions in different time periods. Spatio-temporal graph convolutional networks (STGCNs) are usually adopted to forecast traffic features in a road network. Some STGCN models involves spatial layers first and then temporal layers and some other models involves these layers in a reverse order. This creates an interesting research question on whether the ordering of the spatial layers (or temporal layers) first in an existing STGCN model could improve the forecasting performance. To the best of our knowledge, we are the first to study this interesting research problem, which creates a deep insight as a guideline to the research community on how to design STGCN models. We conducted extensive experiments to study a number of representative STCGN models for this research problem. We found that these models with spatial layers constructed before temporal layers has a higher chance to outperform that with temporal layers constructed first, which suggests the future design principle of STGCN models.

OpenStreetMap (OSM) is a free and openly-editable database of geographic information. Over the years, OSM has evolved into the world's largest open knowledge base of geospatial data, and protecting OSM from the risk of vandalized and falsified information has become paramount to ensuring its continued success. However, despite the increasing usage of OSM and a wide interest in vandalism detection on open knowledge bases such as Wikipedia and Wikidata, OSM has not attracted as much attention from the research community, partially due to a lack of publicly available vandalism corpus. In this paper, we report on the construction of the first OSM vandalism corpus, and release it publicly. We describe a user embedding approach to create OSM user embeddings and add embedding features to a machine learning model to improve vandalism detection in OSM. We validate the model against our vandalism corpus, and observe solid improvements in key metrics. The validated model is deployed into production for vandalism detection on Daylight Map.

Machine learning methods are increasingly leveraged in disparate domains of research. Herein, we describe our curriculum design to introduce undergraduate students to applied research through a series of course assignments and a competition among peers to inspire other educators. We describe the overall course structure and detail how the assignments were tailored to a selected open research question while developing student understanding of machine learning. We outline the lessons learned from this new undergraduate curriculum design and describe how it may be adapted to similar courses. For the selected COVID19-related course-long problem of predicting which drugs might interact with specific proteins, we leveraged state-of-the-art tools for representing drug and protein sequences. We challenged students to develop unique solutions competitive with a current state-of-the-art model using reproducible Notebooks and cloud-based computing resources with the expectation that top-ranking solutions would be used to predict novel druggable targets within the SARS-CoV-2 proteome to possibly treat COVID19 patients. We motivate this curriculum design based on related competition frameworks that have led to notable research advancements and contributed to machine learning pedagogy. From our experience, the top student solutions were ultimately combined using a stacked classifier to create a publishable solution representing an actual research contribution. We highly recommend introducing undergraduate students to open research applications early in their program to encourage them to consider pursuing a career in research.

Delay-sensitive multimedia streaming applications require their data to be delivered before a deadline to be useful. The data transmitted by these applications can usually be partitioned into blocks with different priorities, assigned based on the impact of a block on the Quality of Experience (QoE) if it misses its delivery deadline. Meet their deadline requirements is challenging due to the dynamics of the network and these applications' high demand on network resources. To encourage the research community to address this challenge, we organize the "Meet Deadline Requirements" Grand Challenge at ACM Multimedia 2021. This grand challenge provides a simulation platform onto which the participants can implement their block scheduler and bandwidth estimator and then benchmark against each other using a common set of application traces and network traces.

As Scratch has become one of the most popular educational programming languages, understanding its common programming idioms can benefit both computing educators and learners. This understanding can fine-tune the curricular development to help learners master the fundamentals of writing idiomatic code in their programming pursuits. Unfortunately, the research community's understanding of what constitutes idiomatic Scratch code has been limited. To help bridge this knowledge gap, we systematically identified idioms as based on canonical source code, presented in widely available educational materials. We implemented a tool that automatically detects these idioms to assess their prevalence within a large dataset of over 70K Scratch projects in different experience backgrounds and project categories. Since communal learning and the practice of remixing are one of the cornerstones of the Scratch programming community, we studied the relationship between common programming idioms and remixes. Having analyzed the original projects and their remixes, we observed that different idioms may associate with dissimilar types of code changes. Code changes in remixes are desirable, as they require a meaningful programming effort that spurs the learning process. The ability to substantially change a project in its remixes hinges on the project's code being easy to understand and modify. Our findings suggest that the presence of certain common idioms can indeed positively impact the degree of code changes in remixes. Our findings can help form a foundation of what comprises common Scratch programming idioms, thus benefiting both introductory computing education and Scratch programming tools.

RNA-seq has been widely used in experimental studies and produced a massive amount of data deposited in public databases. New biological insights can be obtained by retrospective analyses of previously published data. However, the barrier to efficiently utilize these data remains high, especially for those who lack bioinformatics skills and computational resources. We present MetazExp (https://bioinfo.njau.edu.cn/metazExp), a database for gene expression and alternative splicing profiles based on 53 615 uniformly processed publicly available RNA-seq samples from 72 metazoan species. The gene expression and alternative splicing profiles can be conveniently queried by gene IDs, symbols, functional terms and sequence similarity. Users can flexibly customize experimental groups to perform differential and specific expression and alternative splicing analyses. A suite of data visualization tools and comprehensive links with external databases allow users to efficiently explore the results and gain insights. In conclusion, MetazExp is a valuable resource for the research community to efficiently utilize the vast public RNA-seq datasets.

In this paper we present a proposed algorithm based on the bunching technique, which can be used sustainably in unobserved distributions. Automating bunching estimation is yet to receive substantial attention from the research community and our method will be one of the firsts attempts in the subject. The purpose of the algorithm is to be flexible, low-maintenance and generalized, when it comes to results; it is structured in several steps, each dependent on the previous. The algorithm concludes with the calculation of the bunching mass, if found, and based on the data distribution around that point, will calculate a significance measure. The algorithm is part of an end-product; an open-source programming language package (R/Python), to be used by the community.

Being the most expensive and most popular cryptocurrency, both the business world and the research community have started to study bitcoin development. However, due to the absence of most government regulation, the price of bitcoin has become uncontrollable, resulting in frequent large fluctuations. Using a dataset from 17 August 2017 to 13 April 2021 the GRU, RNN, and LSTM methods will be compared and implement Grid Search and Random Search to find out which one will do better in this research. Those three methods are considered to be the best method to get a prediction, but it also depends on the model that computer could have. The best result is the GRU with Grid Search method with MAE (Mean Absolute Error) of training 0.0043 and testing about 0.0594.

A vast number of research papers on numerous topics publish every year in different conferences and journals. Thus, it is difficult for new researchers to identify research problems and topics manually, which research community is currently focusing on. Since such research problems and topics help researchers to be updated with new topics in research, it is essential to know trends in research based on topic significance over time. Therefore, in this paper, we propose a method to identify the trends in machine learning research based on significant topics over time automatically. Specifically, we apply a topic coherence model with latent Dirichlet allocation (LDA) to evaluate the optimal number of topics and significant topics for a dataset. The LDA model results in topic proportion over documents where each topic has its probability (i.e., topic weight) related to each document. Subsequently, the topic weights are processed to compute average topic weights per year, trend analysis using rolling mean, topic prevalence per year, and topic proportion per journal title. To evaluate our method, we prepare a new dataset comprising of 21,906 scientific research articles from top six journals in the area of machine learning published from 1988 to 2017. Extensive experimental results on the dataset demonstrate that our technique is efficient, and can help upcoming researchers to explore the research trends and topics in different research areas, say machine learning.

A product recommendation system is aimed at providing an improved shopping experience to the users, thereby increasing the revenues. The text-based product search on most of the online stores is based on the past search history of the user and/or the other characteristics associated to the products, such as the annotated labels, price range, category, description, size, color, and other attributes. Though, this traditional method has adequate performance, however, it is prone to the issues pertaining to improper annotation of the metadata related to the products. Due to the recent advancements in machine learning, product recommendation using machine learning is getting increasing attention. Over the last one decade, tremendous progress in machine learning and computing has paved the way for efficient product recommendation. In this book chapter, we present a critical analysis of product recommendation using the traditional methods. Specifically, the book chapter explores the prospect of product recommendation based on visual similarity using machine learning which is rather a new concept in this domain and is getting increasing attention in research community. The chapter also explores the underlying challenges and discusses the future directives.

Inter-process communication (IPC) has always been the "Achilles heel" of microkernels, determining their overall performance. The entire history of microkernel development is tightly coupled to the debates about IPC, its efficiency, and the bottleneck it creates. The microkernel research community has consistently viewed the idea of message passing as the fundamental foundation of the microkernel, which is considered linked to its design inextricably. We are revising the place of IPC in the microkernel-based multi-server operating system design. We argue that communication can and potentially should be moved out to the userspace making microkernel the first tier resource manager solely. We describe a novel approach to out-of-kernel asynchronous messaging in multi-server microkernel OS and provide preliminary evaluation results to show its benefits.

The analytical description of charts is an exciting and important research area with many applications in academia and industry. Yet, this challenging task has received limited attention from the computational linguistics research community. This paper proposes \textsf{AutoChart}, a large dataset for the analytical description of charts, which aims to encourage more research into this important area. Specifically, we offer a novel framework that generates the charts and their analytical description automatically. We conducted extensive human and machine evaluations on the generated charts and descriptions and demonstrate that the generated texts are informative, coherent, and relevant to the corresponding charts.

In recent years, Smart Farming (SF) and Precision Agriculture (PA) have attracted attention from both the agriculture industry as well as the research community. Altogether, SF and PA aim to help farmers use inputs (such as fertilizers and pesticides) more efficiently through using Internet of Things (IoT) devices, but in doing so, they create new security threats that can defeat this purpose in the absence of adequate awareness and proper countermeasures. A survey on different security-related challenges is required to raise awareness and pave they way for further research in this area. In this paper, we first itemize the security aspects of SF and PA. Next, we review the types of cyber attacks that can violate each of these aspects. Accordingly, we present a taxonomy on cyber-threats to SF and PA on the basis of their relations to different stages of Cyber-Kill Chain (CKC). Among cyber-threats, we choose Advanced Persistent Threats (APTs) for further study. Finally, we studied related risk mitigation strategies and countermeasure, and developed a future road map for further study in this area. This paper’s main contribution is a categorization of security threats within the SF/PA areas and provide a taxonomy of security threats for SF environments so that we may detect the behavior of APT attacks and any other security threat in SF and PA environments.

Machine learning on graphs has been extensively studiedin both academic and industry. However, as the literature on graph learning booms with a vast number of emerging methods and techniques, it becomes increasingly difficult to manually design the optimal machine learning algorithm for different graph-related tasks. To solve this critical challenge, automated machine learning (AutoML) on graphs which combines the strength of graph machine learning and AutoML together, is gaining attentions from the research community. In this tutorial, we discuss AutoML on graphs, primarily focusing on hyper-parameter optimization (HPO) and neural architecture search (NAS) for graph machine learning. We further overview libraries related to automated graph machine learning and in depth discuss AutoGL, the first dedicated open-source library for AutoML on graphs. In the end, we share our insights on future research directions for automated graph machine learning. To the best of our knowledge, this tutorial is the first to systematically and comprehensively review automated machine learning on graphs, possessing a great potential to draw a large amount of interests in the community.

Adversarial learning methods and their applications such as generative adversarial network, adversarial robustness, and security and privacy, have prevailed and revolutionized the research in machine learning and data mining. Their importance has not only been emphasized by the research community but also been widely recognized by the industry and the general public. Continuing the synergies in previous years, this third annual workshop aims to advance this research field. The AdvML'21 workshop consists of three tracks: (i) open-call paper submissions; (ii) invited speakers; and (iii) rising star awards and presentations. The full details about the workshop can be found at https://sites.google.com/view/advml.

COVID‐19 is the disease evoked by a new breed of coronavirus called the severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2). Recently, COVID‐19 has become a pandemic by infecting more than 152 million people in over 216 countries and territories. The exponential increase in the number of infections has rendered traditional diagnosis techniques inefficient. Therefore, many researchers have developed several intelligent techniques, such as deep learning (DL) and machine learning (ML), which can assist the healthcare sector in providing quick and precise COVID‐19 diagnosis. Therefore, this paper provides a comprehensive review of the most recent DL and ML techniques for COVID‐19 diagnosis. The studies are published from December 2019 until April 2021. In general, this paper includes more than 200 studies that have been carefully selected from several publishers, such as IEEE, Springer and Elsevier. We classify the research tracks into two categories: DL and ML and present COVID‐19 public datasets established and extracted from different countries. The measures used to evaluate diagnosis methods are comparatively analysed and proper discussion is provided. In conclusion, for COVID‐19 diagnosing and outbreak prediction, SVM is the most widely used machine learning mechanism, and CNN is the most widely used deep learning mechanism. Accuracy, sensitivity, and specificity are the most widely used measurements in previous studies. Finally, this review paper will guide the research community on the upcoming development of machine learning for COVID‐19 and inspire their works for future development. This review paper will guide the research community on the upcoming development of ML and DL for COVID‐19 and inspire their works for future development.

Despite the multitude of approaches proposed for intrusion detection, cyberattacks are still a timeless issue for the research community and industry as they cause various devastating effects to companies and organisations. There are limited intrusion prediction approaches in the literature, as the main bulk of methods focuses on cyberattack detection rather than prediction, which would allow the defenders (attack’s targets) to restrain/stop the attack. This work aims to identify known DoS and Probe attack patterns at their very beginning. Specifically, we use machine learning algorithms to predict the malicious packets of DoS and Probe attacks, raising the defender’s awareness to act proactively and stop the attack. To the best of our knowledge, this is the first time that time series analysis and machine learning techniques are used to model the intrusion prediction problem effectively. An extensive experimental study confirms the efficacy of the proposed approach according to multiple evaluation measures.

In this paper, we present our progress in pre-training monolingual Transformers for Czech and contribute to the research community by releasing our models for public. The need for such models emerged from our effort to employ Transformers in our language-specific tasks, but we found the performance of the published multilingual models to be very limited. Since the multilingual models are usually pre-trained from 100+ languages, most of low-resourced languages (including Czech) are under-represented in these models. At the same time, there is a huge amount of monolingual training data available in web archives like Common Crawl. We have pre-trained and publicly released two monolingual Czech Transformers and compared them with relevant public models, trained (at least partially) for Czech. The paper presents the Transformers pre-training procedure as well as a comparison of pre-trained models on text classification task from various domains., https://huggingface.co/fav-kky

Adverse drugs effects (ADEs) in children are common and may result in disability and death. However, current pediatric drug safety methods have not gone beyond event surveillance to identify and evaluate potential biological mechanisms. Children undergo an evolutionarily conserved and physiologically dynamic process of growth and maturation that can alter pharmacokinetics and pharmacodynamics. Our hypothesis is that temporal patterns of drug event reporting are reflective of dynamic mechanisms from child growth and development. We generated a database of 460,837 pediatric ADEs using generalized additive models (GAMs) that we have previously shown identify dynamic risk estimates of adverse drug events1. We identified 19,438 significant drug-event risks where drug risks corresponded with physiological development throughout childhood. Our results identified known pediatric drug effects and risk dynamics across child development that were not known previously. For example, we identified significant risk dynamics of montelukast-induced psychiatric disorders, including enriched risk (Odds Ratio 8.77 [2.51, 46.94]) within the second year of life. We developed a data-driven time- series clustering approach resulting in up to 95.2% precision and 97.8% sensitivity for categorizing risk dynamics across development stages for all ADEs including known but previously development-agnostic pediatric drug effects. We found that our real-world evidence may contain biologically-relevant underpinnings as well, where risk dynamics of CYP enzyme substrates were dependent on the enzyme’s expression across childhood. We curated this database for the research community to enable, for the first time, evaluation of real-world hypotheses of adverse drug effects across child growth and development.

During the past few years, video quality assessment (VQA) of user generated content (UGC) has attracted considerable attention in the research community. In this paper, we propose a practical architecture for a versatile video quality model, designed for assessing user generated videos in particular. The proposed architecture is based on our earlier design of two-level video quality model with a convolutional neural network (CNN-TLVQM), with various improvements and re-designed elements. We have built a fast implementation of the proposed model in C++, demonstrating that the model is practical for real-life applications. The implementation of the model has been submitted for evaluation in ICME UGCVQA Challenge in 2021.

Proliferation of “Fake News” and misinformation is resulting in widespread negative social fallout. Scalable Fake News classification techniques for resource poor languages like Hindi are in early stages because of a lack of datasets and lack of robust NLP libraries for these languages. In this exploratory study we curate a dataset of around 13,000 data points of true news articles, and, articles on fake news authored by media organisations which flag fake news. We then use seven ML classification models on this dataset and present the preliminary results. Our results show that concerted efforts need to be made by the research community towards dataset curation and improving the NLP models for resource poor languages in order to make scalable classification systems.

Knowledge about misconceptions is an important element of pedagogical content knowledge. The computing education research community collected a large body of research on misconceptions, using a diverse set of definitions and approaches. Inspired by this prior work, we present an actionable definition of misconceptions, focused on the area most commonly studied: programming and programming languages. We then introduce an organizational structure for collections of programming language misconceptions. We study how existing collections fit our organization, and we present a curated inventory of programming language misconceptions that aims to follow our definition and structure. Our inventory goes beyond traditional programming misconception collections. It connects misconceptions to the authoritative specifications of languages, to places they may be triggered in textbooks, to research papers that discuss them, and it provides support for integrating programming language misconceptions into educational platforms.

The potential harms and drawbacks of automated decision making has become a challenge as data science blends into our lives. In particular, fairness issues with deployed machine learning models have drawn significant attention from the research community. Despite the myriad of algorithmic fairness work in various research communities, in practice data scientists still face many roadblocks in ensuring the fairness of their machine learning models. This is primarily because there does not exist an end-to-end system that guides the users in building a fair machine learning model in a responsible way from model auditing, to model explanation, to bias mitigation. We propose a explorative model building system FairRover for responsible fair model building. FairRover guides users in (1) discovering the potential biases in the model; (2) providing explanation to the discovered biases so as to help users in understanding potential causes of the biases; and (3) mitigating the most important biases selected by the users. Because of the impossibility theorem of fairness, and the well-known trade-off between fairness and accuracy, it is generally impossible to achieve a completely fair and accurate machine learning model. Therefore, this responsible model building process is naturally performed iteratively until a satisfying trade-off is reached. Human users are involved in the loop to make various decisions guided by FairRover. We demonstrate a case study on the Adult Census dataset, which shows how FairRover guides users in iteratively building a fair income prediction model in a responsible way. We discuss the current limitations of FairRover and future work.

Practically anyone can now generate a realistic looking deepfake video. It is clear that the online prevalence of such fake videos will erode the societal trust in video evidence even further. To counter the looming threat, many methods to detect deepfakes were recently proposed by the research community. However, it is still unclear how realistic deep-fake videos are for an average person and whether the algorithms are significantly better than humans at detecting them. Therefore, this paper, presents a subjective study, which, using 60 naive subjects, evaluates how hard it is for humans to see if a video is a deepfake or not. For the study, 120 videos (60 deepfakes and 60 originals) were manually selected from the Facebook database used in Kaggle’s Deepfake Detection Challenge 2020. The results of the subjective evaluation were compared with two state of the art deepfake detection methods, based on Xception and EfficientNet (B4 variant) neural network models pre-trained on two other public databases: Google and Jiqsaw subset from FaceForensics++ and Celeb-DF v2 dataset. The experiments demonstrate that while the human perception is very different from the perception of a machine, both successfully but in different ways are fooled by deepfakes. Specifically, algorithms struggle to detect the deepfake videos that humans find to be very easy to spot.

The proliferation of fake news and its propagation on social media has become a major concern due to its ability to create devastating impacts. Different machine learning approaches have been suggested to detect fake news. However, most of those focused on a specific type of news (such as political) which leads us to the question of dataset-bias of the models used. In this research, we conducted a benchmark study to assess the performance of different applicable machine learning approaches on three different datasets where we accumulated the largest and most diversified one. We explored a number of advanced pre-trained language models for fake news detection along with the traditional and deep learning ones and compared their performances from different aspects for the first time to the best of our knowledge. We find that BERT and similar pre-trained models perform the best for fake news detection, especially with very small dataset. Hence, these models are significantly better option for languages with limited electronic contents, i.e., training data. We also carried out several analysis based on the models' performance, article's topic, article's length, and discussed different lessons learned from them. We believe that this benchmark study will help the research community to explore further and news sites/blogs to select the most appropriate fake news detection method., Comment: 22 pages, 5 figures, to be published in Machine Learning with Applications journal