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

World University root node

Image and Vision Computing

GIFSL-grafting based improved few-shot learning

Road safety is a critical issue worldwide. We believe that mobile devices can play a positive role in this context by detecting dangerous conditions and providing feedback. This paper focuses on a specific problem in developing countries: the stopping behaviour of buses in the vicinity of bus stops. For instance, buses could arrive at a bus stop but continue rolling forward instead of coming to a complete halt, or could stop some distance away from the bus stop, possibly even in the middle of a busy road. Such behaviors put at risk the passengers boarding or alighting the bus, and also the people waiting at a bus stop. We present FullStop, a smartphone-based system that detects safety risks emanating from stopping behavior like the ones listed above. We show that the GPS and inertial sensors are unable to perform the fine-grained detection needed. Therefore, our approach in FullStop is based on the view obtained from looking out to the front of the vehicle using the camera of a smartphone that is mounted on the front windshield. Using optical flow vectors, with several refinements, FullStop running on a smartphone is able to effectively detect various unsafe bus stopping behaviours. © 2002-2012 IEEE.

IEEE Transactions on Mobile Computing

FullStop: A camera-assisted system for characterizing unsafe bus stopping

Zero-shot learning, in spite of its recent popularity, remains an unexplored area for medical image analysis. We introduce a first-of-its-kind generalized zero-shot learning (GZSL) framework that utilizes information from two different imaging modalities (CT and x-ray) for the diagnosis of chest radiographs. Our model makes use of CT radiology reports to create a semantic space consisting of signatures corresponding to different chest diseases and conditions. We introduce a CrOss-Modality Semantic Embedding Ensemble (COMe-SEE) for zero-shot diagnosis of chest x-rays by relating an input x-ray to a signature in the semantic space. The ensemble, designed using a novel semantic saliency preserving autoencoder, utilizes the visual and the semantic saliency to facilitate GZSL. The use of an ensemble not only helps in dealing with noise but also makes our model useful across different datasets. Experiments on two publicly available datasets show that the proposed model can be trained using one dataset and still be applied to data from another source for zero-shot diagnosis of chest x-rays. © 2020, This is a U.S. government work and not under copyright protection in the U.S.; foreign copyright protection may apply.

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

COMe-SEE: Cross-modality Semantic Embedding Ensemble for Generalized Zero-Shot Diagnosis of Chest Radiographs

We introduce a novel problem of localizing all the instances of an object (seen or unseen during training) in a natural image via sketch query. We refer to this problem as sketch-guided object localization. This problem is distinctively different from the traditional sketch-based image retrieval task where the gallery set often contains images with only one object. The sketch-guided object localization proves to be more challenging when we consider the following: (i) the sketches used as queries are abstract representations with little information on the shape and salient attributes of the object, (ii) the sketches have significant variability as they are hand-drawn by a diverse set of untrained human subjects, and (iii) there exists a domain gap between sketch queries and target natural images as these are sampled from very different data distributions. To address the problem of sketch-guided object localization, we propose a novel cross-modal attention scheme that guides the region proposal network (RPN) to generate object proposals relevant to the sketch query. These object proposals are later scored against the query to obtain final localization. Our method is effective with as little as a single sketch query. Moreover, it also generalizes well to object categories not seen during training and is effective in localizing multiple object instances present in the image. Furthermore, we extend our framework to a multi-query setting using novel feature fusion and attention fusion strategies introduced in this paper. The localization performance is evaluated on publicly available object detection benchmarks, viz. MS-COCO and PASCAL-VOC, with sketch queries obtained from ‘Quick, Draw!’. The proposed method significantly outperforms related baselines on both single-query and multi-query localization tasks. © 2020, Springer Nature Switzerland AG.

Sketch-Guided Object Localization in Natural Images

In the Stable Marriage problem, when the preference lists are complete, all agents of the smaller side can be matched. However, this need not be true when preference lists are incomplete. In most real-life situations, where agents participate in the matching market voluntarily and submit their preferences, it is natural to assume that each agent wants to be matched to someone in his/her preference list as opposed to being unmatched. In light of the Rural Hospital Theorem, we have to relax the “no blocking pair” condition for stable matchings in order to match more agents. In this paper, we study the question of matching more agents with fewest possible blocking edges. In particular, the goal is to find a matching whose size exceeds that of a stable matching in the graph by at least t and has at most k blocking edges. We study this question in the realm of parameterized complexity with respect to several natural parameters, k, t, d, where d is the maximum length of a preference list. Unfortunately, the problem remains intractable even for the combined parameter k + t + d. Thus, we extend our study to the local search variant of this problem, in which we search for a matching that not only fulfills each of the above conditions but is “closest”, in terms of its symmetric difference to the given stable matching, and obtain an FPT algorithm. © Sushmita Gupta, Pallavi Jain, Sanjukta Roy, Saket Saurabh, and Meirav Zehavi; licensed under Creative Commons License CC-BY.

Leibniz International Proceedings in Informatics, LIPIcs

On the (parameterized) complexity of almost stable marriage

Discharge voltage of insulation is pivotal in the design of High Voltage systems. In this work, a machine learning algorithm is used to develop a model to predict the discharge characteristics of air. Finite Element Method (FEM) simulations have been performed to extract different electric field and energy features of air gaps in the range 5mm-40mm under lightning impulses of positive polarity. While developing the model, these features along with gap lengths are considered. The features have been used for training a machine learning algorithm based on Gaussian Process Regression (GPR) to develop the model. The results obtained from the model are validated with measured experimental data. A good comparison between the predicted data and the measured data establishes the accuracy of the predicted model. The proposed methodology is compared using different kernel functions. © 2020 IEEE.

Canadian Conference on Electrical and Computer Engineering

Gaussian Process Regression based Model for Prediction of Discharge Voltage of Air Gaps under Positive Polarity Lightning Impulse Voltages

Modern AI systems are reaping the advantage of novel learning methods. With their increasing usage, we are realizing the limitations and shortfalls of these systems. Brittleness to minor adversarial changes in the input data, ability to explain the decisions, address the bias in their training data, high opacity in terms of revealing the lineage of the system, how they were trained and tested, and under which parameters and conditions they can reliably guarantee a certain level of performance, are some of the most prominent limitations. Ensuring the privacy and security of the data, assigning appropriate credits to data sources, and delivering decent outputs are also required features of an AI system. We propose the tutorial on "Trustworthy AI"to address six critical issues in enhancing user and public trust in AI systems, namely: (i) bias and fairness, (ii) explainability, (iii) robust mitigation of adversarial attacks, (iv) improved privacy and security in model building, (v) being decent, and (vi) model attribution, including the right level of credit assignment to the data sources, model architectures, and transparency in lineage. © 2021 ACM.

ACM International Conference Proceeding Series

Trustworthy AI

IIIT-Delhi

Presentation attack detection for face biometrics

Predicting deep learning architecture

A feedback vertex set in a hypergraph H is a set of vertices S such that deleting S from H results in an acyclic hypergraph. Here, deleting a vertex means removing the vertex and all incident hyperedges, and a hypergraph is acyclic if its vertex-edge incidence graph is acyclic. We study the (parameterized complexity of) the Hypergraph Feedback Vertex Set (HFVS) problem: given as input a hypergraph H and an integer k, determine whether H has a feedback vertex set of size at most k. It is easy to see that this problem generalizes the classic Feedback Vertex Set (FVS) problem on graphs. Remarkably, despite the central role of FVS in parameterized algorithms and complexity, the parameterized complexity of a generalization of FVS to hypergraphs has not been studied previously. In this paper, we fill this void. Our main results are as follows HFVS is W[2]-hard (as opposed to FVS, which is fixed parameter tractable). If the input hypergraph is restricted to a linear hypergraph (no two hyperedges intersect in more than one vertex), HFVS admits a randomized algorithm with running time 2O(k3 log k)nO(1). If the input hypergraph is restricted to a d-hypergraph (hyperedges have cardinality at most d), then HFVS admits a deterministic algorithm with running time dO(k)nO(1). The algorithm for linear hypergraphs combines ideas from the randomized algorithm for FVS by Becker et al. [J. Artif. Intell. Res., 2000] with the branching algorithm for Point Line Cover by Langerman and Morin [Discrete &amp; Computational Geometry, 2005]. © Pratibha Choudhary, Lawqueen Kanesh, Daniel Lokshtanov, Fahad Panolan, and Saket Saurabh; licensed under Creative Commons License CC-BY.