Department of Computer Science & Engineering

&ldquo;IIT Jodhpur is committed to impart quality education to facilitate the creation of qualified, competent and responsible human resources to meet the emerging technological challenges of the world. It is working towards creating an equitable learning platform that is positive, accessible and effective in the face of the modern societal and scientific advancements.

Flexibility in curriculum, in all its academic programs, encompassing elements of teaching, research, creativity, innovation and entrepreneurship is the most unique feature of IIT Jodhpur. The Department of Science and Technology (DST), under the National Mission on Cyber Physical Systems, has identified IIT Jodhpur as a Technology &amp; Innovation Hub in the area of Augmented Reality and Virtual Reality. The institute has also been nominated as the coordinator for the Jodhpur City and Knowledge Cluster by the Office of Principal Scientific Adviser, Government of India.

The sprawling 852-acre IIT Jodhpur campus is a majestic tribute to the deep-rooted diverse cultural heritage of the region. The campus is an architectural masterpiece, which is slated to become an international exemplar of sustainability with net-zero energy, and advanced water and waste management strategies.&rdquo;

IIT Jodhpur

The task of gait-based subject recognition (SR) in ubiquitous sensor environments has become popular due to its wide range of applications in biometric authentication and smart home products. In recent times, a significant amount of work has been done in SR using supervised learning algorithms on datasets having high modality. However, the process of annotation for SR is difficult due to the challenges like privacy and high manual cost, which results in a scarcity of labeled data samples. Also, for the datasets having less modality, the task of SR in a semi-supervised domain is sparsely explored and challenging. In this work, we analyze the effect of these two factors (sparse labels and low modality) which are critical for SR in ubiquitous data. We select two datasets of ubiquitous data that are relatively unexplored in the context of SR. The datasets, namely OPPOR-TUNITY and Smartphone used to perform SR using conventional supervised learning algorithms to benchmark the results. Then we perform extensive experimentation to analyze the effect of the aforementioned factors over the task of SR by studying the variations in classification accuracies. Next, we propose our semi-supervised framework for SR based on the concept of pseudo labels to counter the adverse effects of low modality and lack of labels. Experimental results show that our approach offers up to 77% and 98% accuracy on the Smartphone and OPPORTUNITY dataset respectively. © 2020 IEEE.

World University root node

2020 International Conference on COMmunication Systems and NETworkS, COMSNETS 2020

Semi-supervised Subject Recognition through Pseudo Label Generation in Ubiquitous Sensor Data

Influence maximization (IM) is an interesting study in the domain of social and complex networks which has gained widespread importance in recent years due to the growth in viral marketing and targeted advertisements through the use of online social networks. There have been several enriching contributions which focus on efficient algorithms for IM; a fundamental question remains as to how can the IM computations be done efficiently with better performance that can aid evaluations in real time. In this work, we present a novel mechanism of IM computation, using two case studies HBUTA and HSSA, that utilizes all the available computing hardware present in a current-generation high-end computing system. We present techniques for efficiently partitioning work, computations of IM in a distributed manner, consolidation, and addressing challenges towards ensuring maximum coverage of the input graph. We see that with our initial implementations, we get a maximum of 29.05% and 35.67% improvement in performance due to our HBUTA and HSSA algorithms, respectively, when used on a graph consisting of 1.6 million vertices and 30 million edges. © 2019, Springer Science+Business Media, LLC, part of Springer Nature.

Journal of Supercomputing

Accelerating influence maximization using heterogeneous algorithms

Behavioral analysis (BA) on ubiquitous sensor data is the task of finding the latent distribution of features for modeling user-specific characteristics. These characteristics, in turn, can be used for a number of tasks including resource management, power efficiency, and smart home applications. In recent years, the employment of topic models for BA has been found to successfully extract the dynamics of the sensed data. Topic modeling is popularly performed on text data for mining inherent topics. The task of finding the latent topics in textual data is done in an unsupervised manner. In this work we propose a novel clustering technique for BA which can find hidden routines in ubiquitous data and also captures the pattern in the routines. Our approach efficiently works on high dimensional data for BA without performing any computationally expensive reduction operations. We evaluate three different techniques namely LDA, the Non-negative Matrix Factorization (NMF) and the Probabilistic Latent Semantic Analysis (PLSA) for comparative study. We have analyzed the efficiency of the methods by using performance indices like perplexity and silhouette on three real-world ubiquitous sensor datasets namely, the Intel Lab Data, Kyoto Data, and MERL data. Through rigorous experiments, we achieve silhouette scores of 0.7049 over the Intel Lab dataset, 0.6547 over the Kyoto dataset and 0.8312 over the MERL dataset for clustering. © 2020 IEEE.

Behavior Analysis through Routine Cluster Discovery in Ubiquitous Sensor Data

Driver Stress and Behavior prediction is a significant feature of the Advanced Driver Assistance System. This system can improve driving safety by alerting the driver to the danger of unsafe or risky driving conditions. In this paper, we analyzed historical trip data to calculate the driving stress and its impact on different driving behavior. We used Long Short-Term Memory Fully Convolutional Network to predict the corresponding stress level of the driver. We further established a relationship between stress and driving behavior and developed an intelligent recommendation system for cab companies to recommend the driver for a subsequent trip. To meet the demand for Artificial Intelligence in the Intelligent Transportation System, we leverage Federated Learning in Vehicular Edge Computing in the proposed system architecture. It enables Road Side Units to do all computing of data on it. The model has been tested on the UAH-DriveSet dataset. We observed that the proposed model predicts the stress with an accuracy of 95% and assists in enhancing the driving quality and experience. © 2020 IEEE.

Proceedings - 2020 IEEE 17th International Conference on Mobile Ad Hoc and Smart Systems, MASS 2020

Vehicular edge computing based driver recommendation system using federated learning

Cloud Computing offers storage resources as well as network and computing resources to the organizations. This eliminates the high infrastructure cost for the organizations that are using these services as they can now dynamically pay for these services, i.e., pay per use model, which is followed by most of the cloud providers. As the organization does not locally host these resources, these are comparatively far easier to manage and use than the traditional infrastructural resources. As a result of these factors, the popularity of cloud computing is increasing continuously. But this transfer of data and applications to the cloud server also creates some challenges. It poses problems that must be dealt with properly to ensure a secure cloud computing environment. As more and more sensitive data is being uploaded on the cloud in the present scenario, the privacy and security concerns associated with the data is continuously increasing. To address this, issue the data is stored on the cloud in the encrypted form. Also, as the amount of data stored is usually tremendous, so an efficient search scheme is also necessary. So here, we deal with two significant aspects of cloud computing: Encryption and Searching. We are proposing a secure and efficient encryption scheme to encrypt the data stored in the cloud as well as the queries along with a multi-keyword search scheme to search over the encrypted cloud data. © 2020 IEEE.

2020 International Wireless Communications and Mobile Computing, IWCMC 2020

Algorithm for Multi Keyword Search over Encrypted Data in Cloud Environment

Authentication of users and devices is one of the security issues on the internet of things (IoT). Traditionally, the authentication system is based on a centralized server, which makes it prone to various security threats and potential attacks; also, the stored data can tamper. In this paper, existing issues are resolved using Blockchain. IoT devices do not have much computational power. Blockchain technology, which uses the Proof-of-Work concept, consumes a lot of energy, which makes it not suitable for IoT devices. A new efficient scheme to reduce the computational energy of the devices is discussed, along with an authentication scheme that uses no centralized servers and makes IoT devices interact directly with Blockchain is designed and is implemented. The smart contract concept is used in this scheme, which provides users the flexibility to design there own rules for authentication. © 2020, Springer Nature Switzerland AG.

Advances in Intelligent Systems and Computing

Efficient Authentication Scheme Using Blockchain in IoT Devices

We introduce PeeledHuman - a novel shape representation of the human body that is robust to self-occlusions. PeeledHuman encodes the human body as a set of Peeled Depth and RGB maps in 2D, obtained by performing raytracing on the 3D body model and extending each ray beyond its first intersection. This formulation allows us to handle self-occlusions efficiently compared to other representations. Given a monocular RGB image, we learn these Peeled maps in an end-to-end generative adversarial fashion using our novel framework - PeelGAN. We train PeelGAN using a 3D Chamfer loss and other 2D losses to generate multiple depth values per-pixel and a corresponding RGB field per-vertex in a dual-branch setup. In our simple non-parametric solution, the generated Peeled Depth maps are back-projected to 3D space to obtain a complete textured 3D shape. The corresponding RGB maps provide vertex-level texture details. We compare our method with current parametric and non-parametric methods in 3D reconstruction and find that we achieve state-of-theart-results. We demonstrate the effectiveness of our representation on publicly available BUFF and MonoPerfCap datasets as well as loose clothing data collected by our calibrated multi-Kinect setup. © 2020 IEEE.

Proceedings - 2020 International Conference on 3D Vision, 3DV 2020

PeeledHuman: Robust Shape Representation for Textured 3D Human Body Reconstruction

Advances in Neural Information Processing Systems

Calibrating CNNs for Lifelong Learning

Current smartphone-based navigation applications fail to provide lane-level information due to poor GPS accuracy. Detecting and tracking a vehicle's lane position on the road assists in lane-level navigation. For instance, it would be important to know whether a vehicle is in the correct lane for safely making a turn, or whether the vehicle's speed is compliant with a lane-specific speed limit. Recent efforts have used road network information and inertial sensors to estimate lane position. While inertial sensors can detect lane shifts over short windows, it would suffer from error accumulation over time. In this article, we present DeepLane, a system that leverages the back camera of a windshield-mounted smartphone to provide an accurate estimate of the vehicle's current lane. We employ a deep learning-based technique to classify the vehicle's lane position. DeepLane does not depend on any infrastructure support such as lane markings and works even when there are no lane markings, a characteristic of many roads in developing regions. We perform extensive evaluation of DeepLane on real-world datasets collected in developed and developing regions. DeepLane can detect a vehicle's lane position with an accuracy of over 90%, and we have implemented DeepLane as an Android app. © 2020 Association for Computing Machinery.

ACM Transactions on Sensor Networks

Driving lane detection on smartphones using deep neural networks

Inspired by the success of the CNN-RNN framework in the image captioning task, several works have explored this in multi-label image annotation with the hope that the RNN followed by a CNN would encode inter-label dependencies better than using a CNN alone. To do so, for each training sample, the earlier methods converted the ground-truth label-set into a sequence of labels based on their frequencies (e.g., rare-to-frequent) for training the RNN. However, since the ground-truth is an unordered set of labels, imposing a fixed and predefined sequence on them does not naturally align with this task. To address this, some of the recent papers have proposed techniques that are capable to train the RNN without feeding the ground-truth labels in a particular sequence/order. However, most of these techniques leave it to the RNN to implicitly choose one sequence for the ground-truth labels corresponding to each sample at the time of training, thus making it inherently biased. In this paper, we address this limitation and propose a novel approach in which the RNN is explicitly forced to learn multiple relevant inter-label dependencies, without the need of feeding the ground-truth in any particular order. Using thorough empirical comparisons, we demonstrate that our approach outperforms several state-of-the-art techniques on two popular datasets (MS-COCO and NUS-WIDE). Additionally, it provides a new perspecitve of looking at an unordered set of labels as equivalent to a collection of different permutations (sequences) of those labels, thus naturally aligning with the image annotation task. Our code is available at: https://github.com/ayushidutta/multi-order-rnn. © 2020, Springer Nature Switzerland AG.