Department of Computer Science and Engineering

Indian Institute of Technology Hyderabad (IITH) is a premier institute of science and technology established in 2008. IITH has been consistently ranked in the&nbsp;<a href="https://iith.ac.in/reports/" rel="noopener noreferrer">top 10 institutes</a>&nbsp;in India for Engineering&nbsp;<a href="https://www.nirfindia.org/2021/EngineeringRanking.html" rel="noopener noreferrer" target="_blank">according to NIRF</a>&nbsp;making it one of the most coveted schools for science and technology in the country.

IITH offers&nbsp;<a href="https://iith.ac.in/academics/under-graduate/">undergraduate programs</a>&nbsp;in all the classical engineering disciplines, applied sciences, design, as well as several modern interdisciplinary areas. Students are given a flexibility to explore a broad set of areas, and potentially pursue a minor or double major in a discipline that is not their own. Students who wish to seek a deeper understanding of their own discipline are strongly encouraged to get involved in cutting-edge research with the help of a faculty to mentor them, and earn an Honors in their own field.

The very foundation of IIT Hyderabad is based on&nbsp;<a href="https://iith.ac.in/research/">research and innovation</a>. The vibrant research culture is evident from the number of patents and publications that IITH produces consistently every year. IITH offers&nbsp;<a href="https://iith.ac.in/academics/programmes-offered/">graduate programs</a>&nbsp;at both a masters, and a doctoral level, in several diverse areas. There are separate programs for technology, design, science, and liberal arts. The&nbsp;<a href="https://iith.ac.in/academics/post-graduate/">MTech program</a>&nbsp;is offered in various modes and durations to cater to the ever-growing need of postgraduate level professionals in the country.

IITH encourages and supports&nbsp;<a href="https://iic-iith.netlify.app/index.html" rel="noopener noreferrer" target="_blank">innovation and entrepreneurship</a>&nbsp;at every stage. The minor program in entrepreneurship is open to all students of IITH, and many of the courses are offered by industrialists who are entrepreneurs themselves. There are various&nbsp;<a href="https://iith.ac.in/research/centres-incubators/">innovation centers and incubators</a>&nbsp;that promote entrepreneurship and help mentor young innovators. IITH has been very successful in building tie-ups with leading academic institutions around the globe.

IITH enjoys a very special relationship with Japanese universities and industries that goes beyond academic and research collaborations. In fact, some of the iconic buildings in IITH campus will carry the signature of Japanese architecture.

IITH creates a unique holistic ecosystem for education that offers interactive learning, a very flexible academic structure, cutting-edge research, strong industry collaboration, and entrepreneurship. This is an environment which enables students and faculty to translate their dreams into realities.

IIT Hyderabad

The contribution of video content in mobile data traffic is predicted to increase over the coming years. If the network is not managed efficiently, this huge amount of data will increase the congestion in the cellular network. Caching at the edge of the cellular network is one of the prominent solutions to mitigate this issue. By deploying Mobile Edge Computing (MEC) at eNodeB (eNB), distributed edge caching can be realized in the cellular network. In this work, we propose a consolidated caching scheme with cache splitting and trans-rating in MEC network. Cache consolidation uses the collaborative capability of MEC servers to avoid replication of videos in the MEC cache network thus increasing the caching capacity. Without replication, access delay may increase. To maintain low access delay, we split the cache storage into two logical parts; full video cache, and initial video cache. Full video cache stores the complete videos without any replication in the MEC cache network and initial video cache stores only initial segments of the videos. Initial segments of the videos are replicated in the cache network to reduce the access delay. Extensive simulations results show improvement in initial access delay, cache hit ratio, and external traffic load compared to the existing solutions. © 2017 IEEE.

World University root node

11th IEEE International Conference on Advanced Networks and Telecommunications Systems, ANTS 2017

Consolidated caching with cache splitting and trans-rating in mobile edge computing networks

As the Internet of Things (IoT) develops and expands, management of IoT communications becomes a major challenge. A large number of IoT devices may be installed anywhere end users wish, then left unattended and be misused to attack others. In edge networks, it is difficult to properly prevent undesired communication without knowledge of the properties of an IoT service and its devices. In this paper, we argue that application service providers, developers, and network operators should 1) verify and know the authenticity of IoT services, devices, and their communications, and 2) prevent unwanted traffic from IoT devices in a trustworthy, scalable, and distributed manner. This paper proposes a Trust List that represents the distribution of trust among IoT-related stakeholders and provides autonomous enforcement of IoT traffic management at the edge networks by integrating blockchains and Software-Defined Networking (SDN). The principle of Trust List is automating the process of doubting, verifying, and trusting IoT services and devices to effectively prevent attacks and abuses. The proof of concept implementation and experiment of the Trust List using both public and private blockchains reveal its good practice and suggest studies for realistic deployment. © 2018 IEEE.

IEEE World Forum on Internet of Things, WF-IoT 2018 - Proceedings

Trust list: Internet-wide and distributed IoT traffic management using blockchain and SDN

The emergence of the IoT, 5G and different modes of computing has introduced a new demand to tailor-made networks to support a wide spectrum of applications. Even though virtualizing networks and applying QoS to these networks are crucial, it is always challenging to achieve QoS, high acceptance ratio, and cost effectiveness on provisioning virtual networks given the constrained resource of the underlying network. This paper introduces Delay Constraint Optimum Bandwidth Tree (DCOBT), which effectively satisfies the QoS requirement of virtual networks in terms of both end-to-end delay and bandwidth. We propose the QoS-aware Resource Provisioning (QRP) algorithm to determine DCOBT with less bandwidth consumption and superior load balancing. Using SDN as a key platform to implement QoS-aware virtual networks, this paper further proposes flow rule reduction using Destination Label Forwarding (DLF) to provision more virtual networks with less Ternary Content-Addressable Memory (TCAM) consumption. The evaluation results proved significant contribution on different aspects of resource provisioning for QoS-aware virtual networks with improved availability, scalability, and cost effectiveness. © 2018 IEEE.

IEEE/IFIP Network Operations and Management Symposium: Cognitive Management in a Cyber World, NOMS 2018

Effective resource provisioning for QoS-aware virtual networks in SDN

Recent years have seen a growing interest in understanding neural networks from an optimization perspective. It is understood now that converging to low-cost local minima is sucient for such models to become eective in practice. However, in this work, we propose a new hypothesis based on recent theoretical ndings and empirical studies that neural network models actually converge to saddle points with high degeneracy. Our ndings from this work are new, and can have a signicant impact on the development of gradient descent based methods for training neural networks. We validated our hypotheses using an extensive experimental evaluation on standard datasets such as MNIST and CIFAR-10, and also showed that recent eorts that attempt to escape saddles nally converge to saddles with high degeneracy, which we dene as ‘good saddles’. We also veried the famous Wigner’s Semicircle Law in our experimental results. © 2018 Copyright held by the owner/author(s).

ACM International Conference Proceeding Series

Are saddles good enough for neural networks

With recent advances in the field of computer vision and especially deep learning, many fully connected and convolutional neural networks have been trained to achieve state-of-the-art performance on a wide variety of tasks such as speech recognition, image classification and natural language processing. For classification tasks however, most of these deep learning models employ the softmax activation function for prediction and minimize cross-entropy loss. In contrast, we demonstrate a consistent advantage by replacing the softmax layer by a set of binary SVM classifiers organized in a tree or DAG (Directed Acyclic Graph) structure. The idea is to not treat the multiclass classification problem as a whole but to break it down into smaller binary problems where each classifier acts as an expert by focusing on differentiating between only two classes and thus improves the overall accuracy. Furthermore, by arranging the classifiers in a DAG structure, we later also show how it is possible to further improve the performance of the binary classifiers by learning more discriminative features through the same deep network. We validated the proposed methodology on two benchmark datasets, and the results corroborated our claim. © 2018 Elsevier B.V.

Pattern Recognition Letters

Improving multiclass classification by deep networks using DAGSVM and Triplet Loss

Given a host graph G and a pattern graph H, the induced subgraph isomorphism problem is to decide whether G contains an induced subgraph that is isomorphic to H. We study the time complexity of induced subgraph isomorphism problems when the pattern graph is fixed. Nešetřil and Poljak gave an O(nkω) time algorithm that decides the induced subgraph isomorphism problem for any 3k vertex pattern graph (the universal algorithm), where ω is the matrix multiplication exponent. Improvements are not known for any infinite pattern family. Algorithms faster than the universal algorithm are known only for a finite number of pattern graphs. In this paper, we show that there exists infinitely many pattern graphs for which the induced subgraph isomorphism problem has algorithms faster than the universal algorithm. Our algorithm works by reducing the pattern detection problem into a multilinear term detection problem on special classes of polynomials called graph pattern polynomials. We show that many of the existing algorithms including the universal algorithm can also be described in terms of such a reduction. We formalize this class of algorithms by defining graph pattern polynomial families and defining a notion of reduction between these polynomial families. The reduction also allows us to argue about relative hardness of various graph pattern detection problems within this framework. We show that solving the induced subgraph isomorphism for any pattern graph that contains a k-clique is at least as hard detecting k-cliques. An equivalent theorem is not known in the general case. In the full version of this paper, we obtain new algorithms for P5 and C5 that are optimal under reasonable hardness assumptions. We also use this method to derive new combinatorial algorithms – algorithms that do not use fast matrix multiplication – for paths and cycles. We also show why graph homomorphisms play a major role in algorithms for subgraph isomorphism problems. Using this, we show that the arithmetic circuit complexity of the graph homomorphism polynomial for Kk − e (The k-clique with an edge removed) is related to the complexity of many subgraph isomorphism problems. This generalizes and unifies many existing results. © Markus Bläser, Balagopal Komarath, and Karteek Sreenivasaiah.

Leibniz International Proceedings in Informatics, LIPIcs

Graph pattern polynomials

The problem of constructing hazard-free Boolean circuits dates back to the 1940s and is an important problem in circuit design. Our main lower-bound result unconditionally shows the existence of functions whose circuit complexity is polynomially bounded while every hazard-free implementation is provably of exponential size. Previous lower bounds on the hazard-free complexity were only valid for depth 2 circuits. The same proof method yields that every subcubic implementation of Boolean matrix multiplication must have hazards. These results follow from a crucial structural insight: Hazard-free complexity is a natural generalization of monotone complexity to all (not necessarily monotone) Boolean functions. Thus, we can apply known monotone complexity lower bounds to find lower bounds on the hazard-free complexity. We also lift these methods from the monotone setting to prove exponential hazard-free complexity lower bounds for non-monotone functions. As our main upper-bound result we show how to efficiently convert a Boolean circuit into a bounded-bit hazard-free circuit with only a polynomially large blow-up in the number of gates. Previously, the best known method yielded exponentially large circuits in the worst case, so our algorithm gives an exponential improvement. As a side result we establish the NP-completeness of several hazard detection problems. © 2018 Association for Computing Machinery.

Proceedings of the Annual ACM Symposium on Theory of Computing

On the complexity of hazard-Free circuits

Message from the AADDA program co-chairs

This paper proposes an unsupervised debate stance classification algorithm. In other words, finding the side a post author is taking in an online debate. Stance detection has a complementary role in information retrieval, opinion mining, text summarization, etc. Existing stance detection techniques are not able to effectively handle two challenges: Determine whether a given post is a debate or not? If the post is a debate on a given topic, correctly classify the side that the post author is taking. In this paper, we propose techniques that addresses both the above issues. Compared to existing technique, our technique gives 30% improvement in detection of whether a post is a debate or not. Our technique is able to find the side that an author is taking in a debate by 10% higher F1 score compared to existing work. We achieve this improvement by using new syntactic rules, better aspect popularity detection, co-reference resolution, and a novel integer linear programming model to solve the problem. © 2018 Association for Computing Machinery.

Unsupervised stance classification in online debates

Online debate sites act as a popular platform for users to express and form opinions. In this paper, we propose a novel unsupervised approach to perform stance classification of two-sided online debate posts. We propose the use of word embeddings to address the problem of identifying the preferred target of each aspect. We also use word embeddings to train a supervised classifier for selecting only target related aspects. The aspect-target preference information is used to model the stance classification task as an integer linear programming problem. The classifier gives an average aspect classification accuracy of 84% on multiple datasets. Our word embedding based stance classification approach gives 19.80% higher user stance classification accuracy (F1-score) compared to the existing methods. Our results suggest that the use of word embeddings improves accuracy and enables us to perform stance classification without the need for external domain-specific information. © Springer Nature Switzerland AG 2018.