Identification Of Actions Research Articles

Efficient Fingerprinting Attack on Web Applications: An Adaptive Symbolization Approach

Website fingerprinting is valuable for many security solutions as it provides insights into applications that are active on the network. Unfortunately, the existing techniques primarily focus on fingerprinting individual webpages instead of webpage transitions. However, it is a common scenario for users to follow hyperlinks to carry out their actions. In this paper, an adaptive symbolization method based on packet distribution information is proposed to represent network traffic. The Profile Hidden Markov Model (PHMM exploits positional information contained in network traffic sequences and is sensitive to webpage transitional information) is used to construct users’ action patterns. We also construct user role models to represent different kinds of users and apply them to our web application identification framework to uncover more information. The experimental results demonstrate that compared to the equal interval and K-means symbolization algorithms, the adaptive symbolization method retains the maximum amount of information and is less time-consuming. The PHMM-based user action identification method has higher accuracy than the existing traditional classifiers do.

AI versus Human: Rethinking the Role of Agent Knowledge in Consumers’ Coping Mechanism Related to Influencer Marketing

Despite the exponential growth of influencer marketing, little research has examined how consumers’ coping mechanism differs with two types of influencers (AI vs. human). Drawing on the persuasion knowledge model, action identification theory, and construal-level theory, we redefined agent knowledge as a core belief that shapes different autonomy perceptions of the influencers. Employing two single-factor between-subject experiments, Study 1 revealed that consumers perceived AI (vs. human) as a heteronomous agent, which guides a low-level construal devoid of perceived superordinate intentions. Building on these findings, Study 2 showed that consumers’ agent knowledge of AI (vs. human) negatively affected attitudinal and behavioral changes, being sequentially mediated by source credibility and perceived persuasion effectiveness, respectively. Theoretical and practical implications are discussed.

Self-Assessment of General Education and Resource Teachers to Work in The Paradigm of Inclusive Education

The article analyzes the qualitative results obtained from a study of pedagogical specialists' self-assessment of their professional comfort working in an inclusive environment – 151 teachers and 150 resource teachers. The work of pedagogical specialists today is in the process of implementation and development of a new normative system, and tracking their evaluation and self-assessment is informative not only for their satisfaction, but also for the quality of normative provision of the process of education of children and students with different abilities. On the basis of the analysis, a system of conclusions based on the subjective selfassessments is made in the direction of: the need for targeted support of both general education and resource teachers in three directions: 1) logistical through the development of ready-to-use methods for identification and orientation of the necessary actions for the first group and unified validated diagnostic methods for the second group; 2) taking into account the specificities of educational institutions, where there are different but essential needs for effective work, some of which can be addressed locally; 3) receiving regular feedback for the purpose of improving the regulatory framework.

OH-EpiCap: a semi-quantitative tool for the evaluation of One Health epidemiological surveillance capacities and capabilities.

Although international health agencies encourage the development of One Health (OH) surveillance, many systems remain mostly compartmentalized, with limited collaborations among sectors and disciplines. In the framework of the OH European Joint Programme "MATRIX" project, a generic evaluation tool called OH-EpiCap has been developed to enable individual institutes/governments to characterize, assess and monitor their own OH epidemiological surveillance capacities and capabilities. The tool is organized around three dimensions: organization, operational activities, and impact of the OH surveillance system; each dimension is then divided into four targets, each including four indicators. A semi-quantitative questionnaire enables the scoring of each indicator, with four levels according to the degree of satisfaction in the studied OH surveillance system. The evaluation is conducted by a panel of surveillance representatives (during a half-day workshop or with a back-and-forth process to reach a consensus). An R Shiny-based web application facilitates implementation of the evaluation and visualization of the results, and includes a benchmarking option. The tool was piloted on several foodborne hazards (i.e., Salmonella, Campylobacter, Listeria), emerging threats (e.g., antimicrobial resistance) and other zoonotic hazards (psittacosis) in multiple European countries in 2022. These case studies showed that the OH-EpiCap tool supports the tracing of strengths and weaknesses in epidemiological capacities and the identification of concrete and direct actions to improve collaborative activities at all steps of surveillance. It appears complementary to the existing EU-LabCap tool, designed to assess the capacity and capability of European microbiology laboratories. In addition, it provides opportunity to reinforce trust between surveillance stakeholders from across the system and to build a good foundation for a professional network for further collaboration.

Action identification with fusion of BERT and 3DCNN for smart home systems

Action identification for smart home ecosystem has experienced a rising interest among relevant fields. Nevertheless, thoroughly understanding context and movement in video, which is the key component for an effective action identification system and a visual-based smart home ecosystem, remains a challenge. Recently, 3DCNN architectures including SlowFast, have proved to be the proper solution for this issue due to ability to filter spatiotemporal features in video. Additionally, the growth in Natural Language Processing, particularly BERT, has shown the potential of utilizing the Attention Mechanism for extracting temporal relationship of movement throughout a video. We introduce a novel fusing approach aiming to fully exploit the temporal relationship within spatiotemporal features produced by 3DCNN and enhance the efficiency of action identification system for smart home surveillance. The proposed method consists of two models: SlowFast with two distinct ResNet-50 backbones for spatiotemporal features extraction and BERT for aggregating temporal relationship. These models are sequentially stacked using two varieties of fusing techniques: early-ensemble and late-ensemble, constructing a robust unified system with the capability to compete against state-of-the-art modalities. This paper also contribute a new Kinetics51 dataset for our experiment on these mentioned models. The proposed method provides positively high top-1 accuracy for the following datasets: specifically, HMDB-51 dataset, which contains 51 variety of actions, achieves 78.05% using 10-crop testing; Kinetics51, our derived dataset from Kinetics400, achieves 82.53% using 10-crop testing. The performance on challenging action identification datasets HMDB51 and our Kinetics51 has shown the robustness of the mentioned fusing technique applied on two proposed models.

Using modern virtual reality techniques to perform analysis of ITER ECH EL port cell maintenance

In order to maintain an ITER Port Plug after failure, the Port Cell needs to be cleared to allow for a transport cask to access the Port Plug and transfer it for maintenance to the Hot Cell Complex. The Port Cell clearing process is a complex, manual and machine assisted procedure, taking out the equipment blocking cask access in several stages.For the Equatorial Launcher (EL) of the Electron Cyclotron Heating (ECH) system, which is located in Equatorial Port Cell #14 of the ITER tokamak, the Port Cell clearing operation includes the removal of 24 Ex-vessel Waveguides, a complex set of CCWS (component cooling) and PHTS (primary cooling) lines, several manifolds, and other ancillaries, like the Service Vacuum System (SVS) to monitor the torus vacuum tightness.This paper shows how modern Virtual Reality (VR) simulation techniques, in combination with a state of the art, high-end game-PC environment and Head Mounted Display, can be used as a tool to visualize and analyse the maintenance procedures, to predict the time to repair, and to assist in minimizing radiation exposure to take in to consideration the ALARA principles.Special emphasis is put on the verification and validation of manual and machine-assisted maintenance procedures and the identification of tooling requirements. The possibility to simulate logistics and repair actions in an early stage of the design process allows for the identification of those maintenance actions that require dedicated tests or the development of dedicated tools.The added value of using modern VR techniques to analyse the maintenance scenarios are:•Immersive & intuitive – very little training required to experience the process first-hand•Quick and accurate assessment of maintenance scenarios, based on the actual CATIA models•The ability to assess in detail geometrically very complex maintenance sequencing•Less need for expensive hardware mock ups

Identification and molecular mechanism of action of antibacterial peptides from Flavourzyme-hydrolyzed yak casein against Staphylococcus aureus

Antibacterial peptides can be released from yak milk casein. To date, the amino acid sequences and mechanism of action of yak casein-derived antibacterial peptides remain unknown. The current study identified antibacterial peptides from yak casein and their molecular mechanism of action. Our results showed that yak α-casein, β-casein, and κ-casein could be effectively hydrolyzed by Flavourzyme (Solarbio Science and Technology Co. Ltd.), and the 2-h hydrolysate showed the highest antibacterial rate of 43.07 ± 2.59% against Staphylococcus aureus. The 1,000 to 3,000 Da fraction accounted for 23.61% of the 2-h hydrolysate and had an antibacterial rate of 62.64 ± 4.40%. Three novel peptides with antibacterial activity were identified from this fraction, and the β-casein-derived peptide APKHKEMPFPKYP showed the strongest antibacterial effect (half-maximal inhibitory concentration = 0.397 mg/mL). Molecular docking predicted that APKHKEMPFPKYP interacted with 2 important enzymes of Staph. aureus, dihydrofolate reductase and DNA gyrase, through hydrophobic, hydrogen bonding, salt bridge, and π-π stacking interactions. Our findings suggest that the yak casein-derived peptides may serve as a potential source of natural preservatives to inhibit Staph. aureus.

Isolation, identification, and mode of action of antibacterial peptides derived from egg yolk hydrolysate

Egg yolk is a coproduct of egg white processing. The protein hydrolysis of egg yolks to exhibit antimicrobial activity is a strategy for its valorization. The objective of this study is to fractionate antibacterial peptides from pepsin-hydrolyzed egg yolks using flash chromatography. In addition, the mode of actions of the fractionated peptides were elucidated and plausible antibacterial peptides were reported. The fraction 6 (F6) obtained from a C18-flash column exhibited antibacterial activity against Staphylococcus aureus ATCC 29213 and Salmonella typhimurium TISTR 292 at minimal inhibitory concentration (MIC) values of 0.5 to 1 mmol/L (Leucine equivalent). The fractionated peptides induced DNA leakage as monitored by 260 nm. Propidium iodide and SYTO9 staining observed under a confocal microscope suggested the disintegration of cell membranes. Synchrotron-based Fourier-transform infrared spectroscopy analysis revealed that the egg yolk peptides at 1×MIC induced an alteration of phospholipids at cell membranes and modified conformation of intracellular proteins and nucleic acids. Scanning electron microscopy revealed obvious cell ruptures when S. aureus was treated at 1×MIC for 4 h, whereas damage of cell membranes and leakage of intracellular components were also observed for the transmission electron microscopy. Egg yolk peptides showed no hemolytic activity in human erythrocytes at concentrations up to 4 mmol/L. Peptide identification by LC-MS/MS revealed 3 cationic and 10 anionic peptides with 100% sequence similarity to apolipoprotein-B of Gallus gallus with hydrophobicity ranging from 27 to 75%. The identified peptide KGGDLGLFEPTL exhibited the highest antibacterial activity toward S. aureus at MIC of 2 mmol/L. Peptides derived from egg yolk hydrolysate present significant potential as antistaphylococcal agents for food and/or pharmaceutical application.

Abstract 4884: Targeted investigational oncology agents (IOA) in the NCI60: a phenotypic systems-based resource

Abstract The NCI60 human tumor cell line panel is a useful tool in the discovery and development of new anticancer agents. The publicly available cell-line characterization and compound screening data from the NCI60 screen have contributed to identifying cellular mechanisms of potential new anticancer agents. Mean-graph sensitivity/resistance patterns in the NCI60 screen serve as a fingerprint for molecular target identification and mechanism of action (MOA). A new NCI60 resource was developed based on screening of 175 FDA-approved oncology drugs (AOD) plus >825 investigational oncology agents (IOA), representing >250 therapeutic targets and MOAs. Compounds targeting different components in a biochemical pathway tend to show high correlations in their GI50 patterns through COMPARE analysis forming clusters with similar NCI60 mean-graph patterns. COMPARE evaluation of compounds that target components in the PI3K/AKT/mTOR pathway form correlation clusters linked by both target and pathway. Only 3 of 15 AKT inhibitors in the IOA set are not found in the connected cluster with a COMPARE correlation of 0.7 including the multi-kinase inhibitors perifosine and AT-13148. The AKT cluster includes both allosteric (MK-2206) and competitive inhibitors highlighting the NCI60 is a functional cell assay. More than 75% of PI3K inhibitors (32/42 agents), including the selective PI3K alpha and PI3K beta inhibitors, form a highly connected cluster at a 0.7 COMPARE correlation. The heat map view showed that >50% of the non-connected PI3K outliers were inactive in the NCI60 assay. The mTOR inhibitors (12/20) form a connected cluster at 0.7 COMPARE correlation, with half of the non-connected mTOR singletons representative of the rapamycin class of inhibitors. The BRAF/MEK/ERK pathway compounds (37) form a highly connected correlation cluster (13/15 BRAF, 14/15 MEK, 8/9 ERK inhibitors) at a 0.75 COMPARE correlation. The NCI60 heat maps for BRAF inhibitors show a consistent pattern of the melanoma cell lines and 2 of the colon cancer lines (COLO 205, HT29). The ERK inhibitors display sensitivity patterns similar to the BRAF agents with additional activity observed against a leukemia (HL-60), ovarian (OVCAR-5) and renal (A498) cell line and 3 other colon cell lines (HCC-298, HCT-116, SW-620). This data analysis resource will be available to the public (https://ioa.cancer.gov). NCI60 compound suppliers can incorporate their test compound(s) data to use the interactive visualization tools with AOD and IOA agents. This project was funded in part with federal funds from the NCI, NIH, under Contract # 75N91019D00024/75N91020F00003. Citation Format: Joel Morris, Mark W. Kunkel, Stephen L. White, Donn G. Wishka, Omar D. Lopez, Lori Bowles, Penny Sellers, Patricia Ramsey, Julie Grams, Tiffany Rohrer, Karen Martin, Thomas Dexheimer, Dmitriy Sonkin, John D. Williams, Jerry M. Collins, James H. Doroshow, Beverly A. Teicher. Targeted investigational oncology agents (IOA) in the NCI60: a phenotypic systems-based resource. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 4884.

Near real-time animal action recognition and classification

In computer vision, identification of actions of an object is considered as a complex and relevant task. When solving the problem, one requires information on the position of key points of the object. Training models that determine the position of key points requires a large amount of data, including information on the position of these key points. Due to the lack of data for training, the paper provides a method for obtaining additional data for training, as well as an algorithm that allows highly accurate recognition of animal actions based on a small number of data. The achieved accuracy of determining the key points positions within a test sample is 92%. Positions of the key points define the action of the object. Various approaches to classifying actions by key points are compared. The accuracy of identifying the action of the object in the image reaches 72.9 %.

Profiling Experiences of Bullying in the Elementary School: The Role of Gender.

The purpose of the present study was to profile bullying behaviors in elementary schools in Saudi Arabia. A secondary purpose was to examine differences in bullying behaviors across gender. Participants were 3867 fourth graders who completed surveys during the TIMSS 2019 survey. An 11-item bullying experience scale was utilized with good internal consistency reliability. Data were analyzed using latent class analysis with Mplus 8.9 to identify profiles of bullying experiences. The results indicated the presence of five profiles with levels of low, medium, and high bullying experiences, as well as two profiles with no cyberbullying experiences and medium high and medium low physical and verbal instances of bullying. Gender effects were highly pronounced, with most maladaptive bullying profiles being predominantly male. It is concluded that physical bullying is mainly occupied by males and the levels of cyberbullying are generally low in the elementary school grades. Implications for educational policy can clearly direct the development of support groups and expert counseling for both bullies and victims, staff training for identification and course of action, and the development of standardized school policies when such incidences occur.

Understanding school food systems to support the development and implementation of food based policies and interventions

BackgroundSchools provide opportunities to improve the quality of children's diet, whilst reducing inequalities in childhood diet and health. Evidence supports whole school approaches, including consistency in food quality, eating culture and food education. However, such approaches are often poorly implemented due to the highly complex environments in which schools operate. We aimed to develop a school food systems map using a systems thinking approach to help identify the key factors influencing primary school children’s dietary choice.MethodsEight workshops were conducted with 80 children (from schools from varying locations (region of England/UK; urban/rural), deprivation levels and prioritisation of school food policies)) and 11 workshops were held with 82 adult stakeholders across the UK (principals, teachers, caterers, school governors, parents, and local and voluntary sector organisations) to identify factors that influence food choice in children across a school day and their inter-relationships. Initial exploratory workshops started with a ‘blank canvas’ using a group model building approach. Later workshops consolidated findings and supported a wider discussion of factors, relationships and influences within the systems map. Strengths of the relationship between factors/nodes were agreed by stakeholders and individually depicted on the map. We facilitated an additional eight interactive, in-person workshops with children to map their activities across a whole school day to enable the production of a journey map which was shared with stakeholders in workshops to facilitate discussion.ResultsThe final ‘CONNECTS-Food’ systems map included 202 factors that were grouped into 27 nodes. Thematic analysis identified four key themes: leadership and curriculum; child food preference; home environment; and school food environment. Network analysis highlighted key factors that influence child diet across a school day, which were largely in keeping with the thematic analysis; including: 'available funds/resources', 'awareness of initiatives and resources', 'child food preference and intake', 'eligibility of free school meals', 'family circumstances and eating behaviours', 'peer/social norms', 'priorities of head teachers and senior leaders'.ConclusionsOur systems map demonstrates the need to consider factors external to schools and their food environments. The map supports the identification of potential actions, interventions and policies to facilitate a systems-wide positive impact on children’s diets.

Anticancer Small-Molecule Agents Targeting Eukaryotic Elongation Factor 1A: State of the Art.

Eukaryotic elongation factor 1A (eEF1A) canonically delivers amino acyl tRNA to the ribosomal A site during the elongation stage of protein biosynthesis. Yet paradoxically, the oncogenic nature of this instrumental protein has long been recognized. Consistently, eEF1A has proven to be targeted by a wide assortment of small molecules with excellent anticancer activity, among which plitidepsin has been granted approval for the treatment of multiple myeloma. Meanwhile, metarrestin is currently under clinical development for metastatic cancers. Bearing these exciting advances in mind, it would be desirable to present a systematic up-to-date account of the title topic, which, to the best of our knowledge, has thus far been unavailable in the literature. The present review summarizes recent advances in eEF1A-targeting anticancer agents, both naturally occurring and synthetically crafted, with regard to their discovery or design, target identification, structure-activity relationship, and mode of action. Their structural diversity and differential eEF1A-targeting mechanisms warrant continuing research in pursuit of curing eEF1A-driven malignancy.

Study of High-Proficiency Chinese EFL Learners’ Strategy Applications in Constructive Feedback

AbstractIn cross-cultural and interlanguage pragmatics, constructive feedback (CF), defined as “the identification of a problematic action and advice on how to change or correct the problem” (Nguyen & Basturkmen, 2010, p. 125), has received little attention. To fill this gap, similarities and differences between native Chinese speakers, native American speakers, and high-proficiency Chinese EFL learners’ CF are explored in this study. In particular, how these learners’ strategy applications in CF differ from those of American and Chinese speakers is examined. Data were collected through discourse completion tests (DCTs) owing to their advantages in controlling social variables and their efficiency in eliciting rich data within a limited time (Leech, 2014). In total, 42 participants were randomly selected from three groups: 14 native Chinese speakers, 14 native American speakers, and 14 high-proficiency Chinese EFL learners. The results revealed significant differences in strategy employment in CF among the three groups, with the highest disparity elicited in hedge strategies. Here, EFL learners resembled American speakers in six of the eight hedge strategies. However, they had the same percentage in terms of compliments as the Chinese speakers. With regard to supportive moves, there were no significant differences among the three groups. As a pioneering investigation, the aim of this study is to call for further research on CF.

A Computational Meta-Learning Inspired Model for Sketch-based Video Retrieval

Objectives: To design and develop an efficient computing framework for sketch-based video retrieval using fine-grained intrinsic computational approach. Methods: The primary method of sketch-based video retrieval adopts multi-stream multi-modality of joint embedding method for improved P-SBVR from improved fine-grained KTH and TSF related dataset. It considers the potential aspects of the computation of significant visual intrinsic appearance details for sketch objects. The extracted appearance and motion-based features are used to train three different CNN baselines under strong and weak supervision. The system also implements a meta-learning model for different supervised settings to attain better performance of sketch-based video retrieval along with a relational module to overcome the problem of overfitting. Findings: The study derives specific sketch sequences from its formulated dataset to compute instance-level query processing for video retrieval. Further, it also addresses the limitations arising in the context of coarse-grained video retrieval models and sketch-based still image retrieval. The aggregated dataset for rich annotation assisted in the experimental simulation. The experimental evaluation with respect to the performance metric evaluates the 3D CNN baselines under strong supervision and weak-supervision where CNN BL-Type-2 attains maximum video retrieval accuracy of 99.96% for triplet grading feature under relational schema. CNN BL-Type-1 attains maximum retrieval accuracy of 97.40% considering the triplet grading features from the improved SBVR. The evaluation metric for the instance level retrieval process also considers true matching of sketches with the videos, it clearly shows that the appropriate appearance and motion based feature selection has enhanced the video retrieval accuracy up to 96.90% with 99.28% accuracy in action identification considering motion stream, 98.17% for appearance module and 98.45% for fusion module. Another important aspect of the proposed research context is that it addresses the problem of cross-modality while executing the simultaneous matching paradigm for visual appearances of the object with its movement appearing on particular video scenes. The experimental outcome showsits comparable effectiveness relative to the existing system of CNN. Novelty: Unlike the conventional system of sketch analysis, which is more focused on static objects or scenes, the presented approach can efficiently compute the important visual intrinsic appearance details of the object of interest from the sketch and then activate the operations for video retrieval. The proposed CNN based learning model with improved P-SBVR dataset attains better computing time for retrieval with are approximately (200, 210 and 214) milliseconds for CNN BL-Type-1, CNN BL-Type-2, CNN BL-Type-3 and comparable with the existing deep learning based SBVR models. Keywords: Sketch Based Video Retrieval; Intrinsic Appearance Details; Meta Learning; Sketch Dataset; Cross Modality Problem

SOP For Identification of Weak Students and Remedial Action to be Taken.

A class comprises of students of different grasping levels and capabilities and hence it becomes a really difficult task for a teacher to teach all of them with the same pace and stipulated time period. The students in a class can be categorized on the basis of their academic excellence and then additional efforts can be taken accordingly for a group of students. Also, extra assignment can be given to fast learners. Broadly, this can be divided into three steps i.e. identifying the factors that directly have an impact on student’s performance, Individual analysis of students as per the institutional framework ( Mentor-mentee, class teacher, Head of department, Grievance cell, Principal), Using these first two steps, how to enhance the student’s performance in all aspects. This paper is a small effort to initiate the process of identifying, analyzing and providing solution for all round development of slow learners and fast learners.

Application of perturbation gene expression profiles in drug discovery—From mechanism of action to quantitative modelling

High dimensional characterization of drug targets, compound effects and disease phenotypes are crucial for increased efficiency of drug discovery. High-throughput gene expression measurements are one of the most frequently used data acquisition methods for such a systems level analysis of biological phenotypes. RNA sequencing allows genome wide quantification of transcript abundances, recently even on the level of single cells. However, the correct, mechanistic interpretation of transcriptomic measurements is complicated by the fact that gene expression changes can be both the cause and the consequence of altered phenotype. Perturbation gene expression profiles, where gene expression is measured after a genetic or chemical perturbation, can help to overcome these problems by directly connecting the causal perturbations to their gene expression consequences. In this Review, we discuss the main large scale perturbation gene expression profile datasets, and their application in the drug discovery process, covering mechanisms of action identification, drug repurposing, pathway activity analysis and quantitative modelling.

Scheduling Closure Periods Is Not an Effective Management Strategy to Reduce Lampenflora in Show Caves

The conversion of wild caves into tourist sites poses serious threats to the conservation of subterranean environments. Among them, the extensive growth of photosynthetic biofilms induced by artificial lighting—the so-called lampenflora—is of particular concern for cave managers. The identification of cost-effective management actions controlling the growth of lampenflora is therefore required to preserve the environmental and touristic values of show caves. By taking advantage of the closure period imposed to contain the COVID-19 pandemic, we tested whether 6 months of cave closure could be an effective strategy to reduce the concentration of photosynthetic biofilms on speleothems in four geographically close Italian show caves. We compared the concentration of the three main microorganism groups composing lampenflora, i.e., cyanobacteria, diatoms, and green algae, measured in September 2020 with values recorded 6 months after the closure, in May 2021. Although slight variations have been observed across the different sampling sessions, we did not detect any significant effect of the closure period on the overall concentration values of lampenflora. Also, we recorded no significant differences in lampenflora concentration after 4 months of regular tourist use, in September 2021. Our results suggest that management practices based on regulating visits to show caves are not effective strategies to reduce lampenflora. Therefore, management practices aiming at a sustainable use of show caves should focus on the active removal of photosynthetic biofilms.

The Dropout of First-Year STEM Students: Is It Worth Looking beyond Academic Achievement?

The expansion of Higher Education increased the diversity of students, with heterogenous characteristics, needs, and values. Institutions, intending to preserve the mission and the transformative potential of the tertiary level of education, are facing and implementing policies and practices that enhance success conditions, persistence, and avoid student dropout, in order to meet the goals for sustainable development of people and societies. The present study aims to analyze the impact of personal and academic variables on students’ academic difficulties in adaption to HE, academic achievement, and dropout among first-year STEM students. From a cohort of STEM first-year students at a Portuguese public university, the participants numbered 1376. Applying the structural equation modelling, the results highlight the effect of the variables age, gender, scholarship, and grade point average on access to higher education, difficulties in adapting to higher education, and dropout decision. Understanding STEM students who have already dropped out can contribute to better identification of institutional actions to prevent and reduce its occurrence, especially in first-year students.

Cascading Pose Features with CNN-LSTM for Multiview Human Action Recognition

Human Action Recognition (HAR) is a branch of computer vision that deals with the identification of human actions at various levels including low level, action level, and interaction level. Previously, a number of HAR algorithms have been proposed based on handcrafted methods for action recognition. However, the handcrafted techniques are inefficient in case of recognizing interaction level actions as they involve complex scenarios. Meanwhile, the traditional deep learning-based approaches take the entire image as an input and later extract volumes of features, which greatly increase the complexity of the systems; hence, resulting in significantly higher computational time and utilization of resources. Therefore, this research focuses on the development of an efficient multi-view interaction level action recognition system using 2D skeleton data with higher accuracy while reducing the computation complexity based on deep learning architecture. The proposed system extracts 2D skeleton data from the dataset using the OpenPose technique. Later, the extracted 2D skeleton features are given as an input directly to the Convolutional Neural Networks and Long Short-Term Memory (CNN-LSTM) architecture for action recognition. To reduce the complexity, instead of passing the whole image, only extracted features are given to the CNN-LSTM architecture, thus eliminating the need for feature extraction. The proposed method was compared with other existing methods, and the outcomes confirm the potential of the proposed technique. The proposed OpenPose-CNNLSTM achieved an accuracy of 94.4% for MCAD (Multi-camera action dataset) and 91.67% for IXMAS (INRIA Xmas Motion Acquisition Sequences). Our proposed method also significantly decreases the computational complexity by reducing the number of inputs features to 50.