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

Several successful adversarial attacks have demonstrated the vulnerabilities of deep learning algorithms. These attacks are detrimental in building deep learning based dependable AI applications. Therefore, it is imperative to build a defense mechanism to protect the integrity of deep learning models. In this paper, we present a novel "defense layer" in a network which aims to block the generation of adversarial noise and prevents an adversarial attack in black-box and gray-box settings. The parameter-free defense layer, when applied to any convolutional network, helps in achieving protection against attacks such as FGSM, L2, Elastic-Net, and DeepFool. Experiments are performed with different CNN architectures, including VGG, ResNet, and DenseNet, on three databases, namely, MNIST, CIFAR-10, and PaSC. The results showcase the efficacy of the proposed defense layer without adding any computational overhead. For example, on the CIFAR-10 database, while the attack can reduce the accuracy of the ResNet-50 model to as low as 6.3%, the proposed "defense layer" retains the original accuracy of 81.32%. © 2020 IEEE.

World University root node

IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops

DNDNet: Reconfiguring CNN for adversarial robustness

The article deals with the problems of motion detection, object recognition, and scene description using deep learning in the framework of granular computing and Z-numbers. Since deep learning is computationally intensive, whereas granular computing, on the other hand, leads to computation gain, a judicious integration of their merits is made so as to make the learning mechanism computationally efficient. Further, it is shown how the concept of z-numbers can be used to quantify the abstraction of semantic information in interpreting a scene, where subjectivity is of major concern, through recognition of its constituting objects. The system, thus developed, involves recognition of both static objects in the background and moving objects in foreground separately. Rough set theoretic granular computing is adopted where rough lower and upper approximations are used in defining object and background models. During deep learning, instead of scanning the entire image pixel by pixel in the convolution layer, we scan only the representative pixel of each granule. This results in a significant gain in computation time. Arbitrary-shaped and sized granules, as expected, perform better than regular-shaped rectangular granules or fixed-sized granules. The method of tracking is able to deal efficiently with various challenging cases, e.g., tracking partially overlapped objects and suddenly appeared objects. Overall, the granulated system shows a balanced trade-off between speed and accuracy as compared to pixel level learning in tracking and recognition. The concept of using Z-numbers, in providing a granulated linguistic description of a scene, is unique. This gives a more natural interpretation of object recognition in terms of certainty toward scene understanding. © 2019, Springer-Verlag London Ltd., part of Springer Nature.

Neural Computing and Applications

Granulated deep learning and Z-numbers in motion detection and object recognition

In today’s time, the number of IoT devices are increasing rapidly. We everyday hear about Amazon echo, Google Mini, Smart watches etc. These devices collect confidential data of a person and as most of these systems follow centralized architecture approach, So most of the data on internet is basically managed by some central authority or organization. Though these organizations have strict policy regarding data misuse or changing of data without consent but one can’t overlook the fact that these organizations have the ability to do so. Blockchain helps solve such problem as it is not managed by single party if somebody tries to change data in his Blockchain then the hash of the particular block will no longer match and the particular Blockchain will become invalid and other Blockchain will still be intact and therefore users data won’t be compromised. But due to high resource requirement, it becomes problem to run complete Blockchain node on all IoT devices mainly on low power or memory devices. In this paper, we have developed decentralized architecture leveraged from Blockchain technology coupled with an alternative centralized cloud architecture which is a classic client/server architecture with an underlying Blockchain at back-end support with smart contract application written in Solidity language at Ethereum platform. We implement this framework and show how this architecture prevent data misuse by using functionality of Blockchain without requiring the all IoT devices to actually run a Blockchain node. For an end-user experience, it will appear to be same as a normal web app and from the developers perspective, the smart contract application will have similar software design to current web apps thus allowing an easy transition for both of centralized and decentralized methods while still retaining the trust of decentralization. © Springer Nature Switzerland AG 2020.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

A Blockchain Based Solution for Securing Data of IoT Devices

A Software-as-a-Service offering provides ready to use solutions for its tenants. Tenants having different quality requirements are deployed on separate dedicated instances. The concerns related to generating variable quality responses using single instance are not explicitly handled in current design approaches. A lack of standardized tactics and design guidelines also make it difficult for an architect to embed these multi-tenant design decisions at the early stage. In this work, we highlight a few domain-independent architectural concerns that can be used to manage multiple heterogeneous tenants on a shared application instance. We derive some quality requirements that are significant in the multi-tenant scenario, related tactics, measurements and analyze the impact on other software product quality attributes. © 2020 IEEE.

Proceedings - 2020 IEEE International Conference on Software Architecture Companion, ICSA-C 2020

Multi-tenant Quality Attributes to Manage Tenants in SaaS Applications

This paper introduce a new variant of the Genetic Algorithm whichis developed to handle multivariable, multi-objective and very high search space optimization problems like the solving system of non-linear equations. It is an integer coded Genetic Algorithm with conventional cross over and mutation but with Inverse algorithm is varying its search space by varying its digit length on every cycle and it does a fine search followed by a coarse search. And its solution to the optimization problem will converge to precise value over the cycles. Every equation of the system is considered as a single minimization objective function. Multiple objectives are converted to a single fitness function by summing their absolute values. Some difficult test functions for optimization and applications are used to evaluate this algorithm. The results prove that this algorithm is capable to produce promising and precise results. © 2020 Venkatesh SS and D. Mishra, published by De Gruyter.

Fulltext

Journal of Intelligent Systems

Variable Search Space Converging Genetic Algorithm for Solving System of Non-linear Equations

Generative adversarial networks (GANs) have shown remarkable success in the generation of data from natural data manifolds such as images. In several scenarios, it is desirable that generated data is well-clustered, especially when there is severe class imbalance. In this paper, we focus on the problem of clustering in the generated space of GANs and uncover its relationship with the characteristics of the latent space. We derive from first principles, the necessary and sufficient conditions needed to achieve faithful clustering in the GAN framework: (i) presence of a multimodal latent space with adjustable priors, (ii) existence of a latent space inversion mechanism and, (iii) imposition of the desired cluster priors on the latent space. We also identify the GAN models in the literature that partially satisfy these conditions and demonstrate the importance of all the components required, through ablative studies on multiple real-world image datasets. Additionally, we describe a procedure to construct a multimodal latent space which facilitates learning of cluster priors with sparse supervision. Codes for our implementation is available at https://github.com/NEMGAN/NEMGAN-P.

IEEE SIGNAL PROCESSING LETTERS

Effect of the Latent Structure on Clustering With GANs

Compressed Sensing (CS) has heralded a new era in cardiac and body Magnetic Resonance (MR) imaging due to the unprecedented acceleration factors achieved within the ambit of existing scanning hardware, by exploiting the inherent sparsity of MR images in a known basis. Latter works have adopted dictionary learning strategies on vectorized image patches to enforce higher degrees of sparsity. However, the enormous computational complexity has confined them to learning from small patches, thereby foregoing any consideration of the expected structure of the MR images from prior knowledge. In order to efficiently exploit the multi-dimensional characteristic of MR data, this work learns 2D separable dictionaries possessing Kronecker structure at low computational costs by adapting the CANDECOMP/PARAFAC (CP) tensor decomposition method on stacked 2D under-sampled MR slices of similar scan type. The proposed method has been observed to obtain superior reconstruction quality in noiseless and noisy acquisition scenarios over the state-of-art. Furthermore, the learned dictionary can jointly reconstruct a stack of distinct 2D under-sampled slices of similar scan type, in significantly reduced running time. © Springer Nature Singapore Pte Ltd 2020.

Communications in Computer and Information Science

Tensor based dictionary learning for compressive sensing MRI reconstruction

Face recognition in the unconstrained environment is an ongoing research challenge. Although several covariates of face recognition such as pose and low resolution have received significant attention, 'disguise' is considered an onerous covariate of face recognition. One of the primary reasons for this is the scarcity of large and representative labeled databases, along with the lack of algorithms that work well for multiple covariates in such environments. In order to address the problem of face recognition in the presence of disguise, the paper proposes an active learning framework termed as A2-LINK. Starting with a face recognition machine-learning model, A2-LINK intelligently selects training samples from the target domain to be labeled and, using hybrid noises such as adversarial noise, fine-tunes a model that works well both in the presence and absence of disguise. Experimental results demonstrate the effectiveness and generalization of the proposed framework on the DFW and DFW2019 datasets with state-of-the-art deep learning featurization models such as LCSSE, ArcFace and DenseNet.