Computer Science and Engineering

<p>Shiv Nadar University is a comprehensive, multidisciplinary, research-focused, and student-centric University offering a full range of academic programs at the undergraduate, postgraduate and doctoral level. With state-of-the-art infrastructure, the University comprises of academic wings, sophisticated labs, international standard sports facilities, amphitheaters, auditorium, conference rooms and smart classrooms. The University&rsquo;s goal is to become internationally recognized for the quality of its research and creative endeavors and their applicability to improving quality of life, generating new insights and expanding the boundaries of human knowledge creativity. Committed to excellence in teaching, research and service, the University aims to serve the higher education needs of India and the world beyond.</p>

Shiv Nadar University

Communication using mediums like video and audio are essential for a lot of professions. In this paper, interaction with real time audio transmission is looked upon using the tools provided by the domains of IoT and Machine Learning. The transport layer protocols - TCP and UDP are examined for audio transmission quality. Further, different Recurrent Neural Networks (RNN) models are examined for their efficiency in predicting music in case of loss of packets during transmission. These predictions fill in the gaps and help prevent any lag in a real time performance/jam where the loss of musical notes can be a hindrance. © 2020 IEEE.

(SoE) School of Engineering

2020 International Conference on COMmunication Systems and NETworkS, COMSNETS 2020

Networked Remote Music Collaboration using Machine Learning

Research in the realm of accident prediction and analysis is mostly confined to the study of road accidents caused by motor vehicle collisions where the event of vehicle fall from high altitude is mostly ignored. An IoT system has been introduced in this work, which uses a budget smartphone and off-the-shelf sensors to accurately detect and notify vehicle fall event. The proposed system is low cost, reliable and can be easily retrofitted to any type of vehicle. It uses the vehicle speed, absolute linear acceleration and altitude to build an SVM classifier based model to detect the fall event. The proposed IoT system is found to be accurate with a MAPE of 1.8\%. © Springer Nature Singapore Pte Ltd. 2020.

Communications in Computer and Information Science

Modeling Vehicle Fall Detection Event Using Internet of Things

Traffic collisions are synonymous with road accidents worldwide. Falling of vehicles from heights is a significant cause of injury, property damage and death every year. This work addresses the research gap in the study of occurrence, classification and reporting of the vehicle falls. It presents an IoT system, which can detect and classify the occurrence of a vehicle fall event with the help of inbuilt sensors of a contemporary smartphone. The proposed system uses vehicular speed, linear acceleration and altitude values of smartphone and connected sensors to build a k-Nearest Neighbor (k-NN) based fall occurrence and classification model to report the occurrence and severity of a vehicle fall-off. The performance of the introduced model is evaluated with the help of metrics such as precision, recall and F1 score. © 2020 IEEE.

World University root node

Vehicle Fall Severity Modeling using IoT and K-Nearest Neighbor Algorithm

Determination of Indoor Air Quality (IAQ) requires measurement and analysis of all the indoor air parameters. Often, the contribution of some parameters is substantial while that of others is meagre. This work aims to develop an IoT system for monitoring the concentrations of indoor air pollutants. Grey Relational Analysis (GRA) is then used to evaluate all the measured indoor air pollutants and select the ones which influence IAQ the most. The selected parameters are then used to build IAQ forecast model using Artificial Neural Networks (ANN). The accuracy of the ANN based forecast model is measured using mean absolute percentage error (MAPE) and coefficient of determination. © 2020 IEEE.

An IoT-based System to Evaluate Indoor Air Pollutants Using Grey Relational Analysis

The Convolutional Neural Networks (CNNs), in domains like computer vision, mostly reduced the need for handcrafted features due to its ability to learn the problem-specific features from the raw input data. However, the selection of dataset-specific CNN architecture, which mostly performed by either experience or expertise is a time-consuming and error-prone process. To automate the process of learning a CNN architecture, this paper attempts at finding the relationship between Fully Connected (FC) layers with some of the characteristics of the datasets. The CNN architectures, and recently datasets also, are categorized as deep, shallow, wide, etc. This paper tries to formalize these terms along with answering the following questions. (i) What is the impact of deeper/shallow architectures on the performance of the CNN w.r.t. FC layers?, (ii) How the deeper/wider datasets influence the performance of CNN w.r.t. FC layers?, and (iii) Which kind of architecture (deeper/shallower) is better suitable for which kind of (deeper/wider) datasets. To address these findings, we have performed experiments with four CNN architectures having different depths. The experiments are conducted by varying the number of FC layers. We used four widely used datasets including CIFAR-10, CIFAR-100, Tiny ImageNet, and CRCHistoPhenotypes to justify our findings in the context of image classification problem. The source code of this work is available at https://github.com/shabbeersh/Impact-of-FC-layers. © 2019

Fulltext

Neurocomputing

Impact of fully connected layers on performance of convolutional neural networks for image classification

The local descriptors have gained wide range of attention due to their enhanced discriminative abilities. It has been proved that the consideration of multi-scale local neighborhood improves the performance of the descriptor, though at the cost of increased dimension. This paper proposes a novel method to construct a local descriptor using multi-scale neighborhood by finding the local directional order among the intensity values at different scales in a particular direction. Local directional order is the multi-radius relationship factor in a particular direction. The proposed local directional order pattern (LDOP) for a particular pixel is computed by finding the relationship between the center pixel and local directional order indexes. It is required to transform the center value into the range of neighboring orders. Finally, the histogram of LDOP is computed over whole image to construct the descriptor. In contrast to the state-of-the-art descriptors, the dimension of the proposed descriptor does not depend upon the number of neighbors involved to compute the order; it only depends upon the number of directions. The introduced descriptor is evaluated over the image retrieval framework and compared with the state-of-the-art descriptors over challenging face databases such as PaSC, LFW, PubFig, FERET, AR, AT&T, and ExtendedYale. The experimental results confirm the superiority and robustness of the LDOP descriptor. © 2019, Springer Science+Business Media, LLC, part of Springer Nature.

Multimedia Tools and Applications

LDOP: local directional order pattern for robust face retrieval

Link prediction in networks has applications in computer science, graph theory, biology, economics, etc. Link prediction is a very well studied problem. Out of all the different versions, link prediction for unipartite graphs has attracted most attention. In this work we focus on link prediction for bipartite graphs that is based on two very important concepts—potential energy and mutual information. In the three step approach; first the bipartite graph is converted into a unipartite graph with the help of a weighted projection, next the potential energy and mutual information between each node pair in the projected graph is computed. Finally, we present Potential Energy-Mutual Information based similarity metric which helps in prediction of potential links. To evaluate the performance of the proposed algorithm four similarity metrics, namely AUC, Precision, Prediction-power and Precision@K were calculated and compared with eleven baseline algorithms. The Experimental results show that the proposed method outperforms the baseline algorithms. © 2020, The Author(s).

Scientific Reports

A potential energy and mutual information based link prediction approach for bipartite networks

There has been exponential growth in the amount of data being generated on a daily basis. Such a huge amount of data creates a need for efficient data storage techniques. Due to the limitations of existing storage media, new storage solutions have always been of interest. There have been recent developments in order to efficiently use synthetic deoxyribonucleic acid (DNA) for information storage. DNA storage has attracted researchers because of its extremely high data storage density, about 1 exabyte/mm3 and long life under easily achievable conditions. This work presents an encoding scheme for DNA-based data storage system with controllable redundancy and reliability, the authors have also talked about the feasibility of the proposed method. The authors have also analysed the proposed algorithm for time and space complexity. The proposed encoding scheme tries to minimise the bases per letter ratio while controlling the redundancy. They have experimented with three different types of data with a value of redundancy as 0.75. In the randomised simulation setup, it was observed that the proposed algorithm was able to correctly retrieve the stored data in our experiments about 94\% of the time. In the situation, where redundancy was increased to 1, the authors were able to retrieve all the information correctly in the proposed experiments. © The Institution of Engineering and Technology 2020.

IET Nanobiotechnology

Encoding scheme for data storage and retrieval on DNA computers

Stem cell research has come into prominence because of its applications in assisted reproductive technology and the treatment of deadly diseases. In teleost fishes, spermatogonial stem cells have been effectively used to produce surrogate gametes and progeny through germ cell transplantation technique. The present study is the first report of an innovative application of stem cell therapy in fish species for revitalising the reproductive competence of senescent individuals. Senescent male goldfish, Carassius auratus, approximately 10 years of age were procured from a fish-breeding farm and were reared locally in the lab for an additional two years. The senescence of the individuals was then evaluated and confirmed using histological analysis, gonadal index assessment, and germ-cell specific vasa gene expression. Analyses revealed absence of spermatogonial cells and other germ cells in the testes of the senescent fish (n = 5). Spermatogonial cells from sexually immature C. auratus male donor were isolated using discontinuous percoll gradients, labelled with the fluorescent dye PKH-26, and transplanted into the gonads of senescent C. auratus males through urogenital papilla. Six months after the transplant, spermatozoa were collected through applying gentle manual pressure on the abdomen and were observed under a microscope. All C. auratus males with the transplant had produced spermatozoa from the transplanted cells. This was confirmed by the retention of PKH-26 in the spermatozoa and diagnostic SSR locus. Gravid C. auratus females were artificially inseminated with the spermatozoa of those senescent males and natural spawning was allowed. As a result viable progeny were produced. These observation suggests that the reproductive competence of senescent male fishes can be revitalised through spermatogonial stem cell therapy to produce functional gametes. © 2020 Majhi et al.

PeerJ

Resumption of donor-origin spermatogenesis in senescent goldfish Carassius auratus (Linnaeus, 1758) following spermatogonial cell therapy

Biomedical image retrieval is a challenging problem due to the varying contrast and size of structures in the images. The approaches for biomedical image retrieval generally rely on the feature descriptors to characterize the images. The feature descriptor of query image is compared with the descriptors of images from the database, to find the best matches. Several hand-crafted feature descriptors have been proposed so far for biomedical image retrieval by exploiting the local relationship of neighboring image pixels. It is observed in the literature that the local bit-plane decoded features are well suited for this retrieval task. Moreover, in recent past, it is also observed that the convolutional neural network-based features such as AlexNet, Vgg16, GoogleNet and ResNet perform well in many computer vision-related tasks. Motivated by the success of the deep learning-based approaches, this paper proposes a local bit-plane decoding-based AlexNet descriptor (LBpDAD) for biomedical image retrieval. The proposed LBpDAD is computed by max-fusing the ReLU operated feature maps of pre-trained AlexNet at a particular layer, obtained from the original and local bit-plane decoded images. The proposed approach is also compared with Vgg16, GoogleNet and ResNet models. The experiments on the proposed method over three benchmark biomedical databases of different modalities such as MRI, CT and microscopic show the efficacy of the proposed descriptor. © 2019, Springer-Verlag London Ltd., part of Springer Nature.