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

Object detection has been at the forefront for higher level vision tasks such as scene understanding and contextual reasoning. Therefore, solving object detection for a large number of visual categories is paramount. Zero-Shot Object Detection (ZSD) - where training data is not available for some of the target classes - provides semantic scalability to object detection and reduces dependence on large amount of annotations, thus enabling a large number of applications in real-life scenarios. In this paper, we propose a novel multi-space approach to solve ZSD where we combine predictions obtained in two different search spaces. We learn the projection of visual features of proposals to the semantic embedding space and class labels in the semantic embedding space to visual space. We predict similarity scores in the individual spaces and combine them. We present promising results on two datasets, PASCAL VOC and MS COCO. We further discuss the problem of hubness and show that our approach alleviates hubness with a performance superior to previously proposed methods. © 2020 IEEE.

World University root node

Proceedings - 2020 IEEE Winter Conference on Applications of Computer Vision, WACV 2020

A multi-space approach to zero-shot object detection

Active learning approaches in computer vision generally involve querying strong labels for data. However, previous works have shown that weak supervision can be effective in training models for vision tasks while greatly reducing annotation costs. Using this knowledge, we propose an adaptive supervision framework for active learning and demonstrate its effectiveness on the task of object detection. Instead of directly querying bounding box annotations (strong labels) for the most informative samples, we first query weak labels and optimize the model. Using a switching condition, the required supervision level can be increased. Our framework requires little to no change in model architecture. Our extensive experiments show that the proposed framework can be used to train good generalizable models with much lesser annotation costs than the state of the art active learning approaches for object detection. © 2019. The copyright of this document resides with its authors.

30th British Machine Vision Conference 2019, BMVC 2019

An adaptive supervision framework for active learning in object detection

In an attempt to address the huge data demand of indoor mobile users and poor signal strength from outdoor base stations to indoor environments, opertaors have started deploying variety of small cells likes Femtos, picos and micro cells. In this work, we used Femtos as small cells. Femto cells are low-cost, low-power consuming cellular base stations which operate only in licensed spectrum and are designed for both outdoor and indoor communication. Although small cells can be used for enhancing network capacity and coverage, arbitrary deployment of large number of small cells can lead to increase in operators expenditure and may create severe interference issues for cell edge users. In this paper, we look into optimal placement of small cell solutions to improve data rates of users in LTE HetNets using Femtos. Besides, these solutions address the main concerns of interference and resource management by proposing mechanisms for optimal placement of Femtos (OptFP, MinNF), dynamic power control and bandwidth allocation in Femtos (SOPC) and dynamic offloading. We provide a comparison of placement solutions and the applicability of each proposal keeping the operators’ revenue in mind. © 2020, Springer Science+Business Media, LLC, part of Springer Nature.

Wireless Personal Communications

Small Cell Planning: Resource Management and Interference Mitigation Mechanisms in LTE HetNets

In the Zero-touch network and service management (ZSM) architecture, devised by ETSI, making predictions on the observed data is among the functions provided by the analytics block of the control loop cycle. Prediction performance depends on several parameters, such as the utilized computational resources, the leveraged prediction techniques, the deployment location of the prediction tools with respect to the data.This paper proposes a Hybrid Forecast Framework (HFF) running at the network edge or in the cloud to provide fore-casting with the performance required by the control loop cycle. Forecasting at the edge might shorten the control loop cycle if resources shall be made available locally where data is collected. However, in general, edge computational resources are less abundant than the cloud ones, thus causing longer time to perform the prediction. On the opposite, forecasting in the cloud might require more time for the data to reach the utilized tools but more computational resources could be exploited. The HFF is based on utilizing traditional time series analysis prediction algorithms to minimize the utilized resources and energy at the edge while it exploits AI/ML tools to make predictions in the cloud.Results show that for short lead time (i.e., the time, in the future, at which the status of the considered parameter is predicted) edge-based prediction exploiting time series analysis provides better accuracy, requires less resources and time (thus energy) than cloud-based prediction. However, if the lead time is long, cloud-based prediction exploiting Artificial Intelligence/Machine Learning (AI/ML) provides better accuracy. Thus, if the lead time is long, it is preferable because the long lead time compensates for the higher time for prediction, mainly due to data transfer. © 2020 IEEE.

2020 European Conference on Networks and Communications, EuCNC 2020

Orchestrating edge- And cloud-based predictive analytics services

The Feedback Vertex Set problem is undoubtedly one of the most well-studied problems in Parameterized Complexity. In this problem, given an undirected graph G and a non-negative integer k, the objective is to test whether there exists a subset S ⊆ V (G) of size at most k such that G − S is a forest. After a long line of improvement, recently, Li and Nederlof [SODA, 2020] designed a randomized algorithm for the problem running in time O?(2.7k)1. In the Parameterized Complexity literature, several problems around Feedback Vertex Set have been studied. Some of these include Independent Feedback Vertex Set (where the set S should be an independent set in G), Almost Forest Deletion and Pseudoforest Deletion. In Pseudoforest Deletion, each connected component in G− S has at most one cycle in it. However, in Almost Forest Deletion, the input is a graph G and non-negative integers k, ` ∈ N, and the objective is to test whether there exists a vertex subset S of size at most k, such that G − S is ` edges away from a forest. In this paper, using the methodology of Li and Nederlof [SODA, 2020], we obtain the current fastest algorithms for all these problems. In particular we obtain following randomized algorithms. 1. Independent Feedback Vertex Set can be solved in time O?(2.7k). 2. Pseudo Forest Deletion can be solved in time O?(2.85k). 3. Almost Forest Deletion can be solved in O?(min{2.85k · 8.54`, 2.7k · 36.61`, 3k · 1.78`}). © Kishen N. Gowda, Aditya Lonkar, Fahad Panolan, Vraj Patel, and Saket Saurabh.

Leibniz International Proceedings in Informatics, LIPIcs

Improved FPT algorithms for deletion to forest-like structures

It is known that problems like Vertex Cover, Feedback Vertex Set and Odd Cycle Transversal are polynomial time solvable in the class of chordal graphs. We consider these problems in a graph that has at most k vertices whose deletion results in a chordal graph, when parameterized by k. While this investigation fits naturally into the recent trend of what are called “structural parameterizations”, here we assume that the deletion set is not given. One method to solve them is to compute a k-sized or an approximate (f(k) sized, for a function f) chordal vertex deletion set and then use the structural properties of the graph to design an algorithm. This method leads to at least kO(k)nO(1) running time when we use the known parameterized or approximation algorithms for finding a k-sized chordal deletion set on an n vertex graph. In this work, we design 2O(k)nO(1) time algorithms for these problems. Our algorithms do not compute a chordal vertex deletion set (or even an approximate solution). Instead, we construct a tree decomposition of the given graph in time 2O(k)nO(1) where each bag is a union of four cliques and O(k) vertices. We then apply standard dynamic programming algorithms over this special tree decomposition. This special tree decomposition can be of independent interest. Our algorithms are, what are sometimes called permissive in the sense that given an integer k, they detect whether the graph has no chordal vertex deletion set of size at most k or output the special tree decomposition and solve the problem. We also show lower bounds for the problems we deal with under the Strong Exponential Time Hypothesis (SETH). © Ashwin Jacob, Fahad Panolan, Venkatesh Raman, and Vibha Sahlot;

Structural parameterizations with modulator oblivion

We consider ℓ1-Rank-r Approximation over GF(2), where for a binary m × n matrix A and a positive integer constant r, one seeks a binary matrix B of rank at most r, minimizing the column-sum norm ||A − B||1. We show that for every ε ∈ (0, 1), there is a randomized (1 + ε)-approximation algorithm for ℓ1-Rank-r Approximation over GF(2) of running time mO(1)nO(24r·ε−4). This is the first polynomial time approximation scheme (PTAS) for this problem. © 2020 Schloss Dagstuhl- Leibniz-Zentrum fur Informatik GmbH, Dagstuhl Publishing. All rights reserved.

Low-rank binary matrix approximation in column-sum norm

In the Split Vertex Deletion (SVD) problem, the input is an n-vertex undirected graph G and a weight function w: V (G) → N, and the objective is to find a minimum weight subset S of vertices such that G − S is a split graph (i.e., there is bipartition of V (G − S) = C ] I such that C is a clique and I is an independent set in G − S). This problem is a special case of 5-Hitting Set and consequently, there is a simple factor 5-approximation algorithm for this. On the negative side, it is easy to show that the problem does not admit a polynomial time (2 − δ)-approximation algorithm, for any fixed δ &gt; 0, unless the Unique Games Conjecture fails. We start by giving a simple quasipolynomial time (nO(log n)) factor 2-approximation algorithm for SVD using the notion of clique-independent set separating collection. Thus, on the one hand SVD admits a factor 2-approximation in quasipolynomial time, and on the other hand this approximation factor cannot be improved assuming UGC. It naturally leads to the following question: Can SVD be 2-approximated in polynomial time? In this work we almost close this gap and prove that for any ε &gt; 0, there is a nO(log 1ε )-time 2(1 + ε)-approximation algorithm. © Daniel Lokshtanov, Pranabendu Misra, Fahad Panolan, Geevarghese Philip, and Saket Saurabh; licensed under Creative Commons License CC-BY 47th International Colloquium on Automata, Languages, and Programming (ICALP 2020).

A (2 + ε)-factor approximation algorithm for split vertex deletion

A tournament is a directed graph T such that every pair of vertices is connected by an arc. A feedback vertex set is a set S of vertices in T such that T − S is acyclic. We consider the Feedback Vertex Set problem in tournaments. Here the input is a tournament T and a weight function w : V (T) → N and the task is to find a feedback vertex set S in T minimizing w(S) = Pv∈S w(v). Rounding optimal solutions to the natural LP-relaxation of this problem yields a simple 3-approximation algorithm. This has been improved to 2.5 by Cai et al. [SICOMP 2000], and subsequently to 7/3 by Mnich et al. [ESA 2016]. In this paper we give the first polynomial time factor 2 approximation algorithm for this problem. Assuming the Unique Games conjecture, this is the best possible approximation ratio achievable in polynomial time. Copyright © 2020 by SIAM

Proceedings of the Annual ACM-SIAM Symposium on Discrete Algorithms

2-approximating feedback vertex set in tournaments

Echocardiography is a widely used and cost-effective medical imaging procedure that is used to diagnose cardiac irregularities. To capture the various chambers of the heart, echocardiography videos are captured from different angles called views to generate standard images/videos. Automatic classification of these views allows for faster diagnosis and analysis. In this work, we propose a representation for echo videos which encapsulates the motion profile of various chambers and valves that helps effective view classification. This variety of motion profiles is captured in a large Gaussian mixture model called universal motion profile model (UMPM). In order to extract only the relevant motion profiles for each view, a factor analysis based decomposition is applied to the means of the UMPM. This results in a low-dimensional representation called motion profile vector (MPV) which captures the distinctive motion signature for a particular view. To evaluate MPVs, a dataset called ECHO 1.0 is introduced which contains around 637 video clips of the four major views: a) parasternal long-axis view (PLAX), b) parasternal short-axis (PSAX), c) apical four-chamber view (A4C), and d) apical two-chamber view (A2C). We demonstrate the efficacy of motion profile-vectors over other spatio-temporal representations. Further, motion profile-vectors can classify even poorly captured videos with high accuracy which shows the robustness of the proposed representation. © 1982-2012 IEEE.