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

High performance of deep neural network based systems have attracted many applications in object recognition and face recognition. However, researchers have also demonstrated them to be highly sensitive to adversarial perturbation and hence, tend to be unreliable and lack robustness. While most of the research on adversarial perturbation focuses on image specific attacks, recently, image-agnostic Universal perturbations are proposed which learn the adversarial pattern over training distribution and have broader impact on real-world security applications. Such adversarial attacks can have compounding effect on face recognition where these visually imperceptible attacks can cause mismatches. To defend against adversarial attacks, sophisticated detection approaches are prevalent but most of the existing approaches do not focus on image-agnostic attacks. In this paper, we present a simple but efficient approach based on pixel values and Principal Component Analysis as features coupled with a Support Vector Machine as the classifier, to detect image-agnostic universal perturbations. We also present evaluation metrics, namely adversarial perturbation class classification error rate, original class classification error rate, and average classification error rate, to estimate the performance of adversarial perturbation detection algorithms. The experimental results on multiple databases and different DNN architectures show that it is indeed not required to build complex detection algorithms; rather simpler approaches can yield higher detection rates and lower error rates for image agnostic adversarial perturbation. © 2018 IEEE.

World University root node

2018 IEEE 9th International Conference on Biometrics Theory, Applications and Systems, BTAS 2018

Are image-agnostic universal adversarial perturbations for face recognition difficult to detect?

Convolutional Neural Networks have provided state-of-the-art results in several computer vision problems. However, due to a large number of parameters in CNNs, they require a large number of training samples which is a limiting factor for small sample size problems. To address this limitation, we propose SSF-CNN which focuses on learning the 'structure' and 'strength' of filters. The structure of the filter is initialized using a dictionary based filter learning algorithm and the strength of the filter is learned using the small sample training data. The architecture provides the flexibility of training with both small and large training databases, and yields good accuracies even with small size training data. The effectiveness of the algorithm is first demonstrated on MNIST, CIFAR10, and NORB databases, with varying number of training samples. The results show that SSF-CNN significantly reduces the number of parameters required for training while providing high accuracies on the test databases. On small sample size problems such as newborn face recognition and Omniglot, it yields state-of-the-art results. Specifically, on the IIITD Newborn Face Database, the results demonstrate improvement in rank-1 identification accuracy by at least 10%. © 2018 IEEE.

Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition

Learning Structure and Strength of CNN Filters for Small Sample Size Training

Presentation attacks such as using a contact lens with a printed pattern or printouts of an iris can be utilized to bypass a biometric security system. The first international iris liveness competition was launched in 2013 in order to assess the performance of presentation attack detection (PAD) algorithms, with a second competition in 2015. This paper presents results of the third competition, LivDet-Iris 2017. Three software-based approaches to Presentation Attack Detection were submitted. Four datasets of live and spoof images were tested with an additional cross-sensor test. New datasets and novel situations of data have resulted in this competition being of a higher difficulty than previous competitions. Anonymous received the best results with a rate of rejected live samples of 3.36% and rate of accepted spoof samples of 14.71%. The results show that even with advances, printed iris attacks as well as patterned contacts lenses are still difficult for software-based systems to detect. Printed iris images were easier to be differentiated from live images in comparison to patterned contact lenses as was also seen in previous competitions. © 2017 IEEE.

IEEE International Joint Conference on Biometrics, IJCB 2017

LivDet iris 2017 - Iris liveness detection competition 2017

Due to the difficulties of radiologists to detect micro-calcification clusters, computer-aided detection (CAD) system is much needed. Many researchers have undertaken the challenge of building an efficient CAD system and several feature extraction methods are being proposed. Most of them extract low- or mid-level features which restrict the accuracy of the overall classification. We observed that high-level features lead to a better diagnosis and convolutional neural network (CNN) is the best-known model to extract high-level features. In this paper, we propose to use a CNN architecture to do both of the feature extraction and classification task. Our proposed network was applied to both MIAS and DDSM databases, and we have achieved accuracy of 99.074 % and 99.267 %, respectively, which we believe that is the best reported so far. © Springer Nature Singapore Pte Ltd. 2018.

Advances in Intelligent Systems and Computing

Classification of breast masses using convolutional neural network as feature extractor and classifier

There is a growing demand for low power error control decoders that are widely used in communication systems. One of the key component in a turbo error control decoder used to decode turbo codes-the Shannon limit approaching code- is a Max-Log-MAP decoder. To scale down the power and area used by the conventional Add-Compare-Select (ACS) module of the Max-Log-MAP decoder, a modified version of the ACS, known as Compare-Add-Select (CSA) has been used in this paper. The scale down is accomplished by reducing the number of operations used in calculating the path metrics of the Max-Log-MAP decoder. The observations assert that for the proposed architecture, there is a 50% reduction in both area and power without compromising the performance when compared to the conventional architecture. © 2018 IEEE.

2018 International Conference on Advances in Computing, Communications and Informatics, ICACCI 2018

Low Power and Area Efficient Max-log -MAP Decoder

Soft biometric modalities have shown their utility in different applications including reducing the search space significantly. This leads to improved recognition performance, reduced computation time, and faster processing of test samples. Some common soft biometric modalities are ethnicity, gender, age, hair color, iris color, presence of facial hair or moles, and markers. This research focuses on performing ethnicity and gender classification on iris images. We present a novel supervised auto-encoder based approach, Deep Class-Encoder, which uses class labels to learn discriminative representation for the given sample by mapping the learned feature vector to its label. The proposed model is evaluated on two datasets each for ethnicity and gender classification. The results obtained using the proposed Deep Class-Encoder demonstrate its effectiveness in comparison to existing approaches and state-of-the-art methods. © 2017 IEEE.

Gender and ethnicity classification of Iris images using deep class-encoder

First edition. Scope and content: "Deep Learning is now ubiquitous with applied machine learning. All of the technology giants (e.g. Google, Microsoft, Apple, etc.) are focusing on deep learning based techniques for data analytics and artificial intelligence. This technology applies quite strongly to biometrics. This book will cover all the topics in deep learning, namely convolutional neural networks, deep belief network and stacked autoenders. The focus will be on the application of these techniques to various biometric modalities: face, iris, palmprint and fingerprints."--Provided by publisher. Cover; Half Title; Title Page; Copyright Page; Table of Contents; Editors; Contributors; 1: Deep Learning: Fundamentals and Beyond; 2: Unconstrained Face Identification and Verification Using Deep Convolutional Features; 3: Deep Siamese Convolutional Neural Networks for Identical Twins and Look-Alike Identification; 4: Tackling the Optimization and Precision Weakness of Deep Cascaded Regression for Facial Key-Point Localization; 5: Learning Deep Metrics for Person Reidentification; 6: Deep Face-Representation Learning for Kinship Verification; 7: Whatâ#x80;#x99;s Hiding in My Deep Features? 8: Stacked Correlation Filters9: Learning Representations for Unconstrained Fingerprint Recognition; 10: Person Identification Using Handwriting Dynamics and Convolutional Neural Networks; 11: Counteracting Presentation Attacks in Face, Fingerprint and Iris Recognition; 12: Fingervein Presentation Attack Detection Using Transferable Features from Deep Convolution Neural Networks; Index.

Deep Learning in Biometrics

IIITD-Delhi

Heterogeneous deep metric learning for cross-modal biometric recognition

A series of dinuclear Zn(II) complexes [Zn2 (L1) (CH3OH)2(SCN) (OAc)](1), [Zn2 (L1) (CH3OH)2(N3)2](2) and [Zn2 (L1) (Cl)2(CH3OH)]·CH3OH (3) have been synthesized by the reaction of compartmental Schiff base ligand (H2L1) [N,N′-Bis(3-ethoxysalicylidenimino)-1,3-diaminopropane] with Zn(OAc)2·2H2O in presence of coligand like KSCN, NaN3 and NaCl respectively. X-ray diffraction analysis revealed that all the complexes are neutral and possess a 4-membered Zn2 (μ2-O)2 ring fastened by the unified coordination action of a doubly deprotonated ligand. In addition, solid state structure of the complexes display extensive intermolecular interaction which has been supported theoretically by Hirshfeld surface analysis with 2D Fingerprint plots. The synthesized Zn(II) metal complexes observed enhancement of luminescence emission compared to the parent Schiff base due to emanating ligand based intraligand (π→π∗) fluorescence. Additionally, Zn(II) metal complexes exhibited considerable antimicrobial potency against some important Gram +ve and Gram -ve bacteria. © 2017 Elsevier B.V.

Department of Chemistry

Journal of Molecular Structure

Coordination of N,O-donor appended Schiff base ligand (H2L1) towards Zinc(II) in presence of pseudohalides: Syntheses, crystal structures, photoluminescence, antimicrobial activities and Hirshfeld surfaces

Let π be a family of graphs. In the classical π-Vertex Deletion problem, given a graph G and a positive integer k, the objective is to check whether there exists a subset S of at most k vertices such that G − S is in π. In this paper, we introduce the conflict free version of this classical problem, namely Conflict Free π-Vertex Deletion (CF-π-VD), and study these problems from the viewpoint of classical and parameterized complexity. In the CF-π-VD problem, given two graphs G and H on the same vertex set and a positive integer k, the objective is to determine whether there exists a set S ⊑ V(G), of size at most k, such that G − S is in π and H[S] is edgeless. Initiating a systematic study of these problems is one of the main conceptual contribution of this work. We obtain several results on the conflict free version of several classical problems. Our first result shows that if π is characterized by a finite family of forbidden induced subgraphs then CF-π-VD is Fixed Parameter Tractable (FPT). Furthermore, we obtain improved algorithms for conflict free version of several well studied problems. Next, we show that if π is characterized by a “well-behaved” infinite family of forbidden induced subgraphs, then CF-π-VD is W[1]-hard. Motivated by this hardness result, we consider the parameterized complexity of CF-π-VD when H is restricted to well studied families of graphs. In particular, we show that the conflict free versions of several well-known problems such as Feedback Vertex Set, Odd Cycle Transversal, Chordal Vertex Deletion and Interval Vertex Deletion are FPT when H belongs to the families of d-degenerate graphs and nowhere dense graphs. © 2018, Springer International Publishing AG, part of Springer Nature.