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

Deep convolutional neural networks (CNN) have evolved as popular machine learning models for image classification during the past few years, due to their ability to learn the problem-specific features directly from the input images. The success of deep learning models solicits architecture engineering rather than hand-engineering the features. However, designing state-of-the-art CNN for a given task remains a non-trivial and challenging task, especially when training data size is less. To address this phenomena, transfer learning has been used as a popularly adopted technique. While transferring the learned knowledge from one task to another, fine-tuning with the target-dependent fully connected (FC) layers generally produces better results over the target task. In this paper, the proposed AutoFCL model attempts to learn the structure of FC layers of a CNN automatically using Bayesian Optimization. To evaluate the performance of the proposed AutoFCL, we utilize five pre-trained CNN models such as VGG-16, ResNet, DenseNet, MobileNet, and NASNetMobile. The experiments are conducted on three benchmark datasets, namely CalTech-101, Oxford-102 Flowers, and UC Merced Land Use datasets. Fine-tuning the newly learned (target-dependent) FC layers leads to state-of-the-art performance, according to the experiments carried out in this research. The proposed AutoFCL method outperforms the existing methods over CalTech-101 and Oxford-102 Flowers datasets by achieving the accuracy of 94.38 % and 98.89 % , respectively. However, our method achieves comparable performance on the UC Merced Land Use dataset with 96.83 % accuracy. The source code of this research is available at https://github.com/shabbeersh/AutoFCL. © 2021, Springer-Verlag London Ltd., part of Springer Nature.

World University root node

(SoE) School of Engineering

Fulltext

Neural Computing and Applications

AutoFCL: automatically tuning fully connected layers for handling small dataset

As the proportion of time spent by humans in indoor environment increases, it becomes challenging to maintain good air quality for healthy and productive life. The need to develop a context aware, reliable system capable of providing real time information and alerts on indoor air quality is addressed in this paper. The proposed Internet of Things (IoT) system serves to collect data, predict ventilation states, and provide alerts and recommendations to the end user. A novel method for determination of ventilation states using three indoor pollutants PM2.5, PM10 and CO is proposed. Multi-level logistic regression is first used to define indoor ventilation states using ventilation rate which is calculated with the help of indoor CO2 concentration. K-NN classification technique then predicts indoor ventilation state with the help of 3 input attributes, PM2.5, PM10 and CO. Context-aware information about indoor environment and current ventilation state is conveyed to the end-user in form of an alert, through a smartphone application. The system is found to determine the poor ventilation state with accuracy, precision, recall and F1 score values of 94.34\%, 0.91, 0.88, and 0.89, respectively. IEEE

IEEE Internet of Things Journal

Context-Aware Monitoring and Control of Ventilation Rate in Indoor Environments Using Internet of Things

Road accidents are a leading cause of death and disability among youth. Contemporary research on accident detection systems is focused on either decreasing the reporting time or improving the accuracy of accident detection. Internet-of-Things (IoT) platforms have been utilized considerably in recent times to reduce the time required for rescue after an accident. This work presents an IoT-based automotive accident detection and classification (ADC) system, which uses the fusion of smartphone's built-in and connected sensors not only to detect but also to report the type of accident. This novel technique improves the rescue efficacy of various emergency services, such as emergency medical services (EMSs), fire stations, towing services, etc., as knowledge about the type of accident is extremely valuable in planning and executing rescue and relief operations. The emergency assistance providers can better equip themselves according to the situation after making an inference about the injuries sustained by the victims and the damage to the vehicle. In this work, three machine learning models based on Naïve Bayes (NB), Gaussian mixture model (GMM), and decision tree (DT) techniques are compared to identify the best ADC model. Five physical parameters related to vehicle movement, i.e., speed, absolute linear acceleration (ALA), change-in-altitude, pitch, and roll, have been used to train and test each candidate ADC model to identify the correct class of accident among collision, rollover, falloff, and no accident. NB-based ADC model is found to be highly accurate with 0.95 mean F1-score. © 2014 IEEE.

An IoT-Based Vehicle Accident Detection and Classification System Using Sensor Fusion

Action recognition in video sequences is an active research problem in Computer Vision. However, no significant efforts have been made for recognizing actions in hazy videos. This paper proposes a novel unified model for action recognition in hazy video using an efficient combination of a Convolutional Neural Network (CNN) for obtaining the dehazed video first, followed by extracting spatial features from each frame, and a deep bidirectional LSTM (DB-LSTM) network for extracting the temporal features during action. First, each frame of the hazy video is fed into the AOD-Net (All-in-One Dehazing Network) model to obtain the clear representation of frames. Next, spatial features are extracted from every sampled dehazed frame (produced by the AOD-Net model) by using a pre-trained VGG-16 architecture, which helps reduce the redundancy and complexity. Finally, the temporal information across the frames are learnt using a DB-LSTM network, where multiple LSTM layers are stacked together in both the forward and backward passes of the network. The proposed unified model is the first attempt to recognize human action in hazy videos. Experimental results on a synthetic hazy video dataset show state-of-the-art performances in recognizing actions. © 2021, Springer Nature Singapore Pte Ltd.

Communications in Computer and Information Science

Action Recognition in Haze Using an Efficient Fusion of Spatial and Temporal Features

Many online comments on social media platforms are hateful, humorous, or sarcastic. The sarcastic nature of these comments (especially the short ones) alters their actual implied sentiments, which leads to misinterpretations by the existing sentiment analysis models. A lot of research has already been done to detect sarcasm in the text using user-based, topical, and conversational information but not much work has been done to use inter-sentence contextual information for detecting the same. This paper proposes a new deep learning architecture that uses a novel Bidirectional Inter-Sentence Contextual Attention mechanism (Bi-ISCA) to capture inter-sentence dependencies for detecting sarcasm in the user-generated short text using only the conversational context. The proposed deep learning model demonstrates the capability to capture explicit, implicit, and contextual incongruous words & phrases responsible for invoking sarcasm. Bi-ISCA generates results comparable to the state-of-the-art on two widely used benchmark datasets for the sarcasm detection task (Reddit and Twitter). To the best of our knowledge, none of the existing models use an inter-sentence contextual attention mechanism to detect sarcasm in the user-generated short text using only conversational context. © 2021 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).

CEUR Workshop Proceedings

Bi-ISCA: Bidirectional inter-sentence contextual attention mechanism for detecting sarcasm in user generated noisy short text

Haze is an aggregation of very fine, widely dispersed, solid and/or liquid particles suspended in the atmosphere. In this paper, we propose an end-to-end network for single image dehazing, which enhances the CycleGAN model by introducing a transformer architecture within the generator, which is specific for haze removal. The proposed model is trained in an unpaired fashion with clear and hazy images altogether and does not require pairs of hazy and corresponding ground-truth clear images. Furthermore, the proposed model does not depend on estimating the parameters of the atmospheric scattering model. Rather, it uses a K-estimation module as the generator's transformer for complete end-to-end modeling. The feature transformer introduced in the proposed generator model transforms the encoded features into desired feature space and then feeds them into the CycleGAN decoder to create a clear image. In the proposed model we further modified the cycle consistency loss to include the SSIM loss along with pixel-wise mean loss to produce a new loss function specific for the reconstruction task, which enhances the performance of the proposed model. The model performs well even on the high-resolution images provided in the NTIRE 2019 challenge dataset for single image dehazing. Further, we perform experiments on NYU-Depth and reside beta datasets. Results of our experiments show the efficacy of the proposed approach compared to the state-of-the-art in removing the haze from the input image. © 2021 Elsevier Inc.

Journal of Visual Communication and Image Representation

Single image dehazing using improved cycleGAN

The National Capital Region (NCR) encircling the capital of India is the one of the most polluted regions in the world. Poor air quality is a cause of a number of diseases and reduction in life span. Particulate matter (PM) is the most significant as well as the most hazardous air pollutant in this region. This work proposes to build models to analyze and forecast PM concentrations at a location in the NCR. The correlation between PM concentrations in different seasons and with meteorological parameters and other air pollutants is studied to determine the most suitable explanatory variables for building the forecast models. The performance of the proposed models is evaluated with the help of variable importance ranking (VIR), partial plots and measures such as mean error, absolute mean error and root mean square error. © 2021, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.

Journal of Ambient Intelligence and Humanized Computing

Prediction and analysis of particulate matter (PM2.5 and PM10) concentrations using machine learning techniques

Placement of items on the shelf space of retail stores significantly impacts the revenue of the retailer. Since customers typically tend to buy sets of items (i.e., itemsets) together, several research efforts have been undertaken towards facilitating itemset placement in retail stores for improving retailer revenue. However, they fail to consider that the time-period of expiry can vary across items i.e., some items expire sooner than others. This leads to loss of opportunity towards improving retailer revenue. Hence, we propose PEAR, which is a Product Expiry-Aware and Revenue-conscious itemset placement scheme for improving retailer revenue. Our key contributions are three-fold. First, we introduce the problem of addressing retail itemset placement when the items can be associated with different time-periods of expiry. Second, we propose the expiry-aware PEAR scheme for efficiently identifying and placing high-revenue itemsets for improving retailer revenue. Third, we conduct a performance study with two real datasets to demonstrate that PEAR is indeed effective in improving retailer revenue w.r.t. a reference scheme. © 2021 IEEE.

2021 IEEE 8th International Conference on Data Science and Advanced Analytics, DSAA 2021

PEAR: A Product Expiry-Aware and Revenue-Conscious Itemset Placement Scheme

Networks are everywhere and their applications in various fields such as computer science, biological science, economics, and chemical engineering attracted attention of many researchers. Many complex systems in the real world can be represented by networks or graphs. Link prediction is one of the most important tasks in network analysis, thus attracting tremendous research interests in the last decades. In this paper, we present a novel algorithm for link prediction that works efficiently for both unipartite graphs and bipartite graphs. Our novel algorithm is based on the concept of eigenvectors and shortest distance between the nodes. We used the Peron–Frobenius theorem of node importance for link prediction. Four metrics, namely AUC, precision, prediction-power, and precision@K, were computed and compared with fourteen baseline algorithms to test the performance of the proposed algorithm. Testing was done on the thirteen datasets, and experimental results show that the proposed method outperforms the baseline algorithm on the basis of four given metrics. © 2021, The Author(s), under exclusive licence to Springer-Verlag GmbH Austria, part of Springer Nature.

Social Network Analysis and Mining

A novel similarity measure for the link prediction in unipartite and bipartite networks

Climate change is a topic that is frequently debated on social media. A vast majority in the debate cite scientific evidence to recognize the existence of a man-made climate change and its impacts on environment as well as society. The opinion of the masses is critical to dealing with various issues arising due to climate change, such as global warming. In this work, we study people’s opinion on climate change and analyze the data to identify the common topics which garner discussion. Our aim is to analyze the dataset, explore the popular belief of a region and then derive the possible explanation in terms of different factors. This analysis could help us in determining the extent to which different factors affect people’s opinion. By building sentiment analysis models, performing topic modelling and using other appropriate technologies, we can visualise the sentiment pattern to understand the factors affecting them. © 2021, Springer Nature Switzerland AG.