Abstract

Answering questions in the context of videos can be helpful in video indexing, video retrieval systems, video summarization, learning management systems and surveillance video analysis. Although there exists a large body of work on visual question answering, work on video question answering (1) is limited to domains like movies, TV shows, gameplay, or human activity, and (2) is mostly based on common sense reasoning. In this paper, we explore a new frontier in video question answering: answering knowledge-based questions in the context of news videos. To this end, we curate a new dataset of ∼12K news videos spanning across ∼156 h with ∼1M multiple-choice question-answer pairs covering 8263 unique entities. We make the dataset publicly available (https://tinyurl.com/videoQAData). Using this dataset, we

Abstract

The COVID-19 pandemic has reformed the teaching-learning processes in engineering education across the globe. Virtual classrooms substituted physical classrooms with the widespread use of online meeting platforms. The proliferation of virtual classrooms not only paved the way for accelerated digital transformation but also brought back some elementary issues in engineering education. Many engineering students face difficulties in comprehending the fundamental concepts in their courses during virtual learning. As real-world engineering solutions depend on conceptual clarity, misconceptions of basic engineering principles need to be taken seriously. If not identified, analysed and corrected with constructive feedback, misconceptions on various engineering topics can create challenging obstacles in learning. Against this backdrop, this research study introduces a novel solution titled Classification of Students Misconceptions in Individualised Learning Environment (C-SMILE). The primary objective of the C-SMILE system is to examine the usefulness of personalised automated feedback to students to enhance their conceptual understanding by pinpointing their misconceptions. Besides, we propose a method by which students’ misconceptions can be effectively classified for every instructional objective in every engineering course using machine learning techniques. Our pilot-study results show that the proposed C-SMILE system can precisely classify students’ misconceptions in engineering education settings.

Abstract

Objectives: Abnormal vaginal discharge (Sayalan al-Rahim) is a common public health problem that significantly disrupts the health-related quality of life (HRQoL). Syndromic management infers the concurrent treatment of two or more infections. Hence, a comparative, single-blind study was planned to determine the efficacy of Acacia (Acacia nilotica Linn.) pod's sitz bath (Abzan) plus vaginal pessary (Farzaja) vs. placebo in abnormal vaginal discharge syndromic management, its associated symptoms, and women's HRQoL

Abstract

Elliptic Curve Cryptography (ECC)-based authenti-cation schemes for the Radio-frequency identification (RFID) environment have emerged as a very safe, secure, and robust candidate of mutual authentication. However, the limited resource availability in a passive tag makes such schemes difficult to realize in practice. Public/private key pair generation, storage, and the computation using those keys are the major bottlenecks towards the faster and efficient design of such schemes. Therefore, we have designed our scheme without using public/private key pairs. We only utilize the Elliptic Curve Discrete Logarithm Problem (ECDLP) property for the realization of our key-less scheme on the secure elliptic curves. Our scheme is primarily aimed at efficient implementation of authentication schemes in Warehouse Management System (WMS), where the data is stored in local servers. This new idea helps to save the memory space in the tags and the server. Moreover, the computation cost also reduces to a great extent in comparison to other schemes. We have compared our scheme with several other schemes and found it efficient in terms of storage cost as well. The simulation result of our scheme with the popular aut

Abstract

Emotion recognition is a complex task that involves understanding and scrutinizing the information depicted by the human body in the form of various physiological signals. Earlier works have used these signals for manually extracting features, such as by using facial landmarks or by using Mel frequency cepstrum coefficient (MFCCs), to classify emotions. But with increasing computational power, neural networks are now widely being used to automate this task. Many state-of-the-art convolutional neural network (CNN) models can be used to extract features and classify emotions. In this research, a quantitative comparison has been done between five state-of-the-art CNN models—DenseNet, ResNet, Inception, VGG16, and

Abstract

The robustness and availability of cloud services are becoming increasingly important as more applications migrate to the cloud. The operations landscape today is more complex, than ever. Site reliability engineers (SREs) are expected to handle more incidents than ever before with shorter service-level agreements (SLAs). By exploiting log, tracing, metric, and network data, Artificial Intelligence for IT Operations (AIOps) enables detection of faults and anomalous issues of services. A wide variety of anomaly detection techniques have been incorporated in various AIOps platforms (e.g. PCA and autoencoder), but they all suffer from false positives. In this paper, we propose an unsupervised approach for persistent anomaly detection on top of the traditional anomaly detection approaches, with the goal of reducing false positives and providing more trustworthy alerting signals. We test our method on both simulated and real-world datasets. Our technique reduces false positive anomalies by at least 28%, resulting in more reliable and trustworthy notifications. CCS CONCEPTS • Computing methodologies → Anomaly detection;. Software and its engineering →Maintaining software. ACM Reference

Abstract

Poststroke and traumatic elbow injuries are the most common cause of elbow sti…ness, which results in loss of functional range of motion (ROM). Various studies support early mobilization of the elbow joint after injury or after surgery to reduce risks of elbow sti…ness development. After hospitalization, patients are required to follow a long-term rehabilitation program during home recovery. Still, most patients do not adhere to their clinical therapy schedule due to either rehabilitation cost, social obligations, negligence, or lack of inspiration. Moreover, the numbers of therapists and assessment equipment are insuŽcient. is study introduces a smart elbow brace (SEB) as a home-based rehabilitation device that reduces regular in-patient rehabilitation costs and therapist workload and motivates patients to comply with the rehabilitation program that enhances the achievement of rehabilitation goals. Our device has two active degrees of freedom (2-DoF) that allow extension, exion, pronation, and supination elbow motions. An extra sliding joint between forearm and wrist is added, which helps dump forces concentration at the elbow joint during extension-exion movement. Mechanical design requirements, motion-tracking systems, and serious game development are described. e feasibility of a proposed SEB device is tested with •ve healthy subjects playing developed serious games with the device. e results show that subjects can attain maximum and minimum angles of exion-extension and pronation-supination motion designed for elbow sti…ness rehabilitation. e SEB device will be bene•cial and be used at home as a complementary tool to support elbow sti…ness rehabilitation during long-term home recovery.

Abstract

Aim of the study: To examine the effect of controlled-cortical impact (CCI), a preclinical model of traumatic brain injury (TBI), on intestinal integrity using a binary classification model of machine learning (ML).Materials and methods: Adult, male C57BL/6 mice were subjected to CCI surgery using a stereotaxic impactor (Impact One™). The rotarod and hot-plate tests were performed to assess the neurological deficits.Results: Mice underwent CCI displayed a remarkable neurological deficit as noticed by decreased latency to fall and lesser paw withdrawal latency in rotarod and hot plate test, respectively. Animals were sacrificed 3 days post-injury (dpi). The colon sections were stained with hematoxylin and eosin (H&E) to integrate with machinery tool-based algorithms. Several stained colon images were captured to build a dataset for ML model to predict the impact of CCI vs sham procedure. The best results were obtained with VGG16 features with SVM RBF kernel and VGG16 features with stacked fully connected layers on top. We achieved a test accuracy of 84% and predicted the disrupted gut permeability and epithelium wall of colon in CCI group as compared to sham-operated mice.Conclusion: We suggest that ML may become an important tool in the development of preclinical TBI model and discovery of newer therapeutics. Keywords: Traumatic brain injury; artificial intelligence; controlled-cortical impact; gut sections; machine learning.

Abstract

Tuberculosis (TB) is a communicable pulmonary disorder and countries with low and middle-income share a higher TB burden as compared to others. The year 2020–2021 universally saw a brutal pandemic in the form of COVID-19, that crushed various lives, health infrastructures, programs, and economies worldwide at an unprecedented speed. The gravity of this estimation gets intensified in systems with limited technological advancements. To assist in the identification of tuberculosis, we propose the ensembling of efficient deep convolutional networks and machine learning algorithms that do not entail heavy computational resources. In this paper, the three of the most efficient deep convolutional networks and machine learning algorithms are employed for resource-effective (low computational and basic Imaging requirements) detection of Tuberculosis. The pivotal features extracted from the deep networks are ensembled and subsequently, the machine learning algorithms are used to identify the images based on the extracted features. The said model underwent k-fold cross-validation and achieved an accuracy of 87.90% and 99.10% with an AUC of 0.94 and 1 respectively in identifying TB infected images from Normal and COVID infected images. Also, the model’s error rate, F-score, and youden’s index values of 0.0093, 0.9901, and 0.9812 for TB versus COVID identification along with the model’s

Abstract

In this paper, we study the three-node Decode-and-Forward (D&F) relay network subject to random and burst packet erasures. The source wishes to transmit an infinite stream of packets to the destination via the relay. The three-node D&F relay network is constrained by a decoding delay of T packets, i.e., the packet transmitted by the source at time i must be decoded by the destination by time i + T . For the individual channels from source to relay and relay to destination, we assume a delay-constrained sliding-window (DCSW) based packet-erasure model that can be viewed as a tractable approximation to the commonly-accepted Gilbert-Elliot channel model. Under the model, any time-window of width w contains either up to a random erasures or else erasure burst of length at most b (≥ a). Thus the source-relay and relay-destination channels are modelled as (a 1 , b 1 , w 1 , T 1 ) and (a 2 , b 2 , w 2 , T 2 ) DCSW channels. We first derive an upper bound on the capacity of the three-node D&F relay network. We then show that the upper bound is tight for the parameter regime: max{b 1 , b 2 } | (T − b 1 − b 2 − max {a 1 , a 2 } + 1) by constructing streaming codes achieving the bound. The code construction requires field size linear in T , and has decoding complexity equivalent to that of decoding an MDS code.