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

In the next-generation radio access network (NG-RAN) defined by 3GPP for the fifth-generation mobile communications, the next-generation NodeB is split into a radio unit (RU), a distributed unit (DU), and a central unit (CU). The RU, DU, and CU are connected through the fronthaul (RU-DU) and midhaul (DU-CU) segments. If the RAN is also a virtualized RAN (VRAN), DU and CU are deployed in virtual machines or containers. Different latency and jitter requirements are demanded on the midhaul according to the distribution of the protocol functions between the DU and CU. This letter shows that in VRAN, the virtualization technologies, the functional split option, and the number of elements deployed in the same computational resource affect the latency budget available for the midhaul. Moreover, it provides an expression for the midhaul allowable latency as a function of the aforementioned parameters. Finally, it shows that the virtualized DUs featuring a lower layer split option shall be deployed not in the same computational resources, where other vDUs are deployed. © 2019 IEEE.

World University root node

Fulltext

IEEE Communications Letters

Impact of Virtualization Technologies on Virtualized RAN Midhaul Latency Budget: A Quantitative Experimental Evaluation

This paper proposes a simple, yet very effective method to localize dominant foreground objects in an image, to pixel-level precision. The proposed method ‘MASON’ (Model-AgnoStic ObjectNess) uses a deep convolutional network to generate category-independent and model-agnostic heat maps for any image. The network is not explicitly trained for the task, and hence, can be used off-the-shelf in tandem with any other network or task. We show that this framework scales to a wide variety of images, and illustrate the effectiveness of MASON in three varied application contexts. © Springer Nature Switzerland AG 2019.

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

MASON: A model agnostic objectness framework

In Second or Foreign Language (SFL) education, a number of studies in applied linguistics have addressed a common issue of how teachers can provide effective feedback to correct learner's erroneous utterances during a classroom hour. Oral Corrective Feedback (OCF) is generally time-consuming and labor-intensive work for teachers. The use of ASR (Automatic Speech Recognition) in SFL education has drawn attention from both teachers and learners to increase the learning effect and efficiency. We designed and integrated Quantitative OCF using Google Cloud Speech-to-Text as a part of the oral assessment using an LMS (Learning Management System) for Japanese SFL courses. The level of learners is a starter's level without any prerequisite knowledge of Japanese language. Preliminary experiments using a total of 214 audio datasets by non-native speakers exhibited that 37.4% of the datasets were recognized properly as Japanese sentences. However, as the remainder of the datasets contains erroneous utterances, characteristics of intonation, or noise, ASR successfully detected word-based errors with high accuracy (82.4%) but low precision (28.1%). Oral assessment employing ASR is highly promising as a complementary system for teachers on partially automating the assessment of audio data from learners with evidence and priority orders as well as significantly reducing teachers' scoring workload and time spent on the most problematic part of the students' speech. While our implementation still requires teachers to double-check, such overhead is small and affordable. © 2019 Association for Computing Machinery.

ACM International Conference Proceeding Series

Employing automatic speech recognition for quantitative oral corrective feedback in Japanese second or foreign language education

In this work, we present a novel meta-learning algorithm that regresses model parameters for novel tasks for which no ground truth is available (zero-shot tasks). In order to adapt to novel zero-shot tasks, our meta-learner learns from the model parameters of known tasks (with ground truth) and the correlation of known tasks to zero-shot tasks. Such intuition finds its foothold in cognitive science, where a subject (human baby) can adapt to a novel concept (depth understanding) by correlating it with old concepts (hand movement or self-motion), without receiving an explicit supervision. We evaluated our model on the Taskonomy dataset, with four tasks as zero-shot: Surface normal, room layout, depth and camera pose estimation. These tasks were chosen based on the data acquisition complexity and the complexity associated with the learning process using a deep network. Our proposed methodolgy outperforms state-of-the-art models (which use ground truth) on each of our zero-shot tasks, showing promise on zero-shot task transfer. We also conducted extensive experiments to study the various choices of our methodology, as well as showed how the proposed method can also be used in transfer learning. To the best of our knowledge, this is the first such effort on zero-shot learning in the task space. © 2019 IEEE.

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

Zero-shot task transfer

Generating an image from its description is a challenging task worth solving because of its numerous practical applications ranging from image editing to virtual reality. All existing methods use one single caption to generate a plausible image. A single caption by itself, can be limited and may not be able to capture the variety of concepts and behavior that would be present in the image. We propose two deep generative models that generate an image by making use of multiple captions describing it. This is achieved by ensuring ‘Cross-Caption Cycle Consistency’ between the multiple captions and the generated image(s). We report quantitative and qualitative results on the standard Caltech-UCSD Birds (CUB) and Oxford-102 Flowers datasets to validate the efficacy of the proposed approach. © 2019 IEEE

Proceedings - 2019 IEEE Winter Conference on Applications of Computer Vision, WACV 2019

C4Synth: Cross-caption cycle-consistent text-to-image synthesis

An undirected graph G is d-degenerate if every subgraph of G has a vertex of degree at most d. By the classical theorem of Erdős and Gallai from 1959, every graph of degeneracy d &gt; 1 contains a cycle of length at least d + 1. The proof of Erdős and Gallai is constructive and can be turned into a polynomial time algorithm constructing a cycle of length at least d + 1. But can we decide in polynomial time whether a graph contains a cycle of length at least d + 2? An easy reduction from Hamiltonian Cycle provides a negative answer to this question: Deciding whether a graph has a cycle of length at least d + 2 is NP-complete. Surprisingly, the complexity of the problem changes drastically when the input graph is 2-connected. In this case we prove that deciding whether G contains a cycle of length at least d + k can be done in time 2O(k)|V (G)|O(1). In other words, deciding whether a 2-connected n-vertex G contains a cycle of length at least d + log n can be done in polynomial time. Similar algorithmic results hold for long paths in graphs. We observe that deciding whether a graph has a path of length at least d + 1 is NP-complete. However, we prove that if graph G is connected, then deciding whether G contains a path of length at least d + k can be done in time 2O(k)nO(1). We complement these results by showing that the choice of degeneracy as the “above guarantee parameterization” is optimal in the following sense: For any ε &gt; 0 it is NP-complete to decide whether a connected (2-connected) graph of degeneracy d has a path (cycle) of length at least (1 + ε)d. © Fedor V. Fomin, Petr A. Golovach, Daniel Lokshtanov, Fahad Panolan, Saket Saurabh, and Meirav Zehavi.

Leibniz International Proceedings in Informatics, LIPIcs

Going far from degeneracy

We report a tool for analysing through statistical model checking, complex dynamical systems on graphs that can be modelled as multi-agent systems. We discuss techniques to leverage the fact that we restrict the tool to dynamics on graphs for performance improvements. The query language that the tool provides is a probabilistic version of bounded linear temporal logic. We also introduce the notion of population sampling on agents for statistical model checking. To the best of our knowledge, this feature has not been reported previously in literature. Finally, we report experimental results on running examples that illustrate our ideas and the utility of the tool. © 2019, Springer Nature Switzerland AG.

DyNeMoC: Statistical Model Checking for Agent Based Systems on Graphs

Massive increase in multimedia data has created a need for effective organization strategy. The multimedia collection is organized based on attributes such as domain, index-terms, content description, owners, etc. Typically, index-term is a prominent attribute for effective video retrieval systems. In this paper, we present a new approach of automatic video tagging referred to as ViTag. Our analysis relies upon various image similarity metrics to automatically extract key-frames. For each key-frame, raw tags are generated by performing reverse image tagging. The final step analyzes raw tags in order to discover hidden semantic information. On a dataset of 103 videos belonging to 13 domains derived from various YouTube categories, we are able to generate tags with 65.51% accuracy. We also rank the generated tags based upon the number of proper nouns present in it. The geometric mean of Reciprocal Rank estimated over the entire collection has been found to be 0.873. © 2019 IEEE.

Proceedings - 2019 IEEE 5th International Conference on Multimedia Big Data, BigMM 2019

ViTag: Automatic video tagging using segmentation and conceptual inference

Word embedding techniques in literature are mostly based on Bag of Words models where words that co-occur with each other are considered to be related. However, it is not necessary for similar or related words to occur in the same context window. In this paper, we propose a new approach to combine different types of resources for training word embeddings. The lexical resources used in this work are Dependency Parse Tree and WordNet. Apart from the co-occurrence information, the use of these additional resources helps us in including the semantic and syntactic information from the text in learning the word representations. The learned representations are evaluated on multiple evaluation tasks like Semantic Textual Similarity, Word Similarity. Results of the experimental analyses highlight the usefulness of the proposed methodology. © 2019 Association for Computing Machinery.

Proceedings of the ACM Symposium on Document Engineering, DocEng 2019

Multi-context information for word representation learning

We propose a compiler driven acceleration of parallel computations on GPUs by exploiting the various special varieties of caches (texture, surface and constant for NVIDIA GPUs). We show that our method obtains better performance for a class of computations when compared with earlier methods that use on-chip shared memory. We provide an end-to-end solution by developing a fully automatic, sound, static framework within a state-of-art source-to-source Polyhedral compiler (PPCG) to exploit these varieties of GPU caches. We use polyhedral model for profitability modeling of the particular variety of GPU caches. We evaluate our implementation on PolyBench/C benchmark kernels and report up to 1. 5x speedups over the current memory mapping strategy used by PPCG compiler. We also consider sample real-world representative kernels: PageRank, DNN layer (LSTM), solvers (Poisson and DWE-FDTD stencil), and show that using the special GPU caches in these programs results in up to 2. 6x speedup over a standard shared memory based implementation. We believe that our contribution is towards automatic exploitation of GPU cache/memory hierarchy as it shows general purpose computing usage of special GPU caches that were originally designed for image processing applications. © 2019 IEEE.