ASISTECH: Assistive Technology Solutions for Mobility & Education of Visually Impaired
Abstract: Inclusion of marginalized communities is considered a major developmental objective for all societies in this millennia. Inclusion of people with disabilities, born or acquired later, constitutes a very different challenge as it cannot be handled just by policies and political will. It requires major technology intervention. In this talk we present the work of ASSISTECH, IIT Delhi in developing solutions aimed at independent mobility and education of visually impaired in India and other low income countries. ASSISTECH has had limited success in disseminating the technology solutions that have been developed in the laboratory. First, we would focus on uniqueness of our Indian context and how these solutions innovatively address our needs including affordability. Further, we also talk about the challenges in translational research and dissemination that is critical for taking any proof-of-concept beyond the walls of a laboratory.
Biography : M. Balakrishnan is a Professor in the Department of Computer Science & Engineering and currently Deputy Director(Strategy & Planning) at IIT Delhi. He obtained his B.E.(Hons.) in Electronics & Electrical Engg. from BITS Pilani in 1977 and Ph.D. from EE Dept. IIT Delhi in 1985. He worked as a Scientist in CARE, IIT Delhi from 1977 to 1985 where he was involved in designing and implementing real-time DSP systems. For the last 30 years, he is involved in teaching and research in the areas of digital systems design, electronic design automation and embedded systems. He has supervised 12 Ph.D. students, 3 MSR students, 173 M.Tech/B.Tech projects and published nearly 110 conference and journal papers. Further, he has held visiting positions in universities in Canada, USA and Germany. At IIT Delhi, he has been the Philips Chair Professor, Head of the Department of Computer Science & Engineering, Dean of Post Graduate Studies & Research, Deputy Director (Faculty) and is currently Deputy Director (Strategy & Planning). At IIT Delhi, he has initiated a number of programmes/activities to promote research, start-ups as well as outreach to the community.
ASSISTECH, a laboratory and research group founded by him along with Prof. P.V.M. Rao (Mech Engg), is involved in developing a number of assistive devices targeted towards mobility and education of the visually impaired. He has been a recipient of three National awards for his work in the disability space. SmartCaneTM is a mobility aid for visually impaired developed by his group and currently it is used by thousands of users in India and other low-income countries.
AI Solutions for the Underserved Billions
While AI has become ubiquitous in the daily lives of most people in developed segments of the world, it’s usage among the billions of poor is practically non-existent. Yet, given that the a primary cause of sustained poverty and hardship is the insufficiency of human expertise and inadequacy of expert human resources, AI may actually provide opportunities to transform the every aspect of the life such as health, agriculture, basic education, infrastructure, and financial inclusion. Wadhwani AI is an independent not-for-profit applied research Institute based in Mumbai founded on this premise and with the goal of developing innovative AI based solutions to address the challenges of the lives of the poor. We are about one year old, and have grown to a team of about 25, working in developing solutions for a few initial use-cases in health and agriculture, specifically in the areas of Maternal and Child health, TB eradication, and Cotton Farming.
In this talk, I will described our approach to these problems, including how we select our problems, the technical aspects of our work, and how we propose to see our work have tangible, sustained and scalable impact in the world. I will also use the talk to share some the learnings from the first year of this endeavor, especially regarding some of the key challenges in doing AI in the social sector and how to address them.
P. Anandan has recently joined the newly formed Wadhwani Institute of Artificial Intelligence as its CEO. Previous to this Anandan was VP for Research at the Adobe Research Lab India (2016-2017) and before that a Distinguished Scientist and Managing Director at Microsoft Research (1997-2014). Anandan was the founding director of Microsoft Research India which he ran from 2005-2014. Prior to this, Anandan was researcher at Sarnoff corporation (1991-1997) and an Assistant Professor of Computer Science at Yale University (1987-1991). His primary research area is Computer vision where he is well known for his fundamental and lasting contributions to the problem of visual motion analysis.
Anandan received his PhD in Computer Science from University of Massachusetts, Amherst in 1987, a Masters in Computer Science from University of Nebraska, Lincoln in 1979 and his BTech in Electrical Engineering from IIT Madras, India in 1977. He is a distinguished alumnus of IIT Madras, and UMass, Amherst and is on the Nebraska Hall of Computing.
Dr. Indrajit Bhattacharya
Discovering Topical Interactions in Text-based Cascades using Hidden Markov Hawkes Processes
Abstract:Social media conversations unfold based on complex interactions between users, topics and time. While recent models have been proposed to capture network strengths between users, users’ topical preferences and temporal patterns between posting and response times, interaction patterns between topics has not been studied. We propose the Hidden Markov Hawkes Process (HMHP) that incorporates topical Markov Chains within Hawkes processes to jointly model topical interactions along with user-user and user-topic patterns. We propose a Gibbs sampling algorithm for HMHP that jointly infers the network strengths, diffusion paths, the topics of the posts as well as the topic-topic interactions. We show using experiments on real and semisynthetic data that HMHP is able to generalize better and recover the network strengths, topics and diffusion paths more accurately than state-of-the-art baselines. More interestingly, HMHP finds insightful interactions between topics in real tweets which no existing model is able to do.
Biography : Dr. Indrajit Bhattacharya works with the TCS Innovation Labs, Kolkata. His webpage can be accessed here.
Making 5G NR a Reality
Affordable healthcare solutions for Ophthalmology and Nephrology
Abstract: Todays healthcare suffers from sever lack of resources and hence causing a steep raise in healthcare costs. In India we have 20,000 ophthalmologists and 1800 nephrologists for 1.3 billion population. Only 10% of the needed population is being attended and the majority of them either have no affordability, more so accessibility and availability of care when needed. Digitizing healthcare can help in greatly overcoming these problems or it is better to say that this is the only feasible solution. Instead of patient go to hospital - hospital go to patient, is the new paradigm which can emerge out of this digitization. With our Digi Health Platform (DHP) is one step towards digitization of healthcare, the talk will focus on two such case studies one is in ophthalmology and the other in chronic disease management for nephrology. This case studies illustrates how we can get accessibility for quality care to the masses, perform task shifting and reduce the cost of expensive resources like health specialists and make care available anywhere anytime.
Biography : Dr. Shyam VasudevaRaois the Managing Director of Forus Health Pvt Ltd.,Director MYMO wireless,Technical Director Maastricht University, Faculty of Health, medicine and life sciences, The Netherlands, Co-founder & DirectorRenalyX health systems, eHealth enablers, Rx Digi Health Platform all in Bangalore. Chief Mentor Centre for Innovation in Medical Electronics, BMS Engineering College. His research work has been on Real time systems and Parallel Computer Architecture from the Indian Institute of Science& has worked with CG Smith as Team Lead, Ericsson General Manager, Tata Consultancy Services Sr Product Manager and Philips Medical Systems. He did his engineering and masters in Electronics from SJCE, Mysore University. His first innovation was a hardware antivirus solution called Knox card. He co invented this in 1989 and this let to a start-up called Knoxware which was one the first to come up with such a unique solution to computer virus protection in a generic way. During his stint at Ericsson, Dr. Shyam was responsible for building a R&D team of more than 100 engineers for new product development activities at Bangalore. After this success, he was on special assignment at their head office in Stockholm, as a Strategic Product Manager for over 2 years. This involved building revenues and profits, through innovative and creative business strategies and extensive customer-centric business analysis. Working as a Director of technology in Philips Innovation Campus, Bangalore, he established the innovation framework for consumer electronics and medical systems division. He exponentially grew the patent portfolio ofboth the above divisions and established working relationships with academic institutions. He has been serving as member Board of Governance and academic council of several reputed institutions like SJCE Mysore, JSS School of Management, JSS ATE Bangalore, PES Institute of technology, ATMA Institute of technology Mysore, East west college of Engineering, Nandi engineering college, Adarsha Institute of technology and board of studies at MSRIT, BMS, BNMIT and VTU and visiting faculty at MIT Manipal, and BMS engineering college. He has won several awards for his thesis at IISc a gold medal and Prof. Badkas best thesis award, His innovation 3nethra, an all in one eye screening device, at Forus has won several Loral’s like DST Locked Martin Gold medal, Perimal Award, Sankalp Award, CNBC TV18 Award, Samsung Innovation Quotient Award, NASCOM Innovation Award and Anjani Mashelkar Inclusive Innovation Award for 2011. In 2016 IOT India Congress awarded IOT Thought Leader. In 2017 one of my innovative product at Forus Neo won the Marico Innovation award. He has published more than 40 technical papers and filed more than 25 international patents with 5 US patent grants.
Quality Assessment of Image Stitching for Virtual Reality
Abstract: Virtual Reality technologies aim to provide a realistic and immersive experience in virtual environments to the users through wide field of view images, which are typically obtained by stitching multiple images with overlapping fields of view. We consider the problem of Quality Assessment (QA) of stitched images for virtual reality applications. Our contributions are two-fold. Firstly, we design a stitched image QA database consisting of 264 stitched images and 6600 human quality ratings. The database consists of a variety of artifacts due to stitching such as blur, ghosting, color and geometric distortions. Secondly, we propose an objective QA algorithm called the Stitched Image Quality Evaluator (SIQE) based on models of natural scene statistics to capture distortions that arise during stitching. We show through extensive experiments that our proposed quality model correlates very well with subjective scores present in the database.
Biography : Rajiv Soundararajan received the B. E. degree in Electrical and Electronics Engineering from Birla Institute of Technology and Science (BITS), Pilani, India in 2006. He received the M. S. and Ph. D. degrees in Electrical and Computer Engineering from The University of Texas at Austin, USA in 2008 and 2012 respectively. Between 2012 and 2015, he was with Qualcomm Research India, Bangalore. He is currently an Assistant Professor at the Indian Institute of Science, Bangalore. He received the 2016 IEEE Circuits and Systems for Video Technology Best Paper Award and the 2017 IEEE Signal Processing Letters Best Paper Award. His research interests are broadly in image and video signal processing, information theory, computer vision and machine learning.
Programmability at low overhead for heterogeneous computing
As the benefits from transistor scaling diminish, an increasingly larger number of applications are making use of domain-specific accelerators, such as general purpose graphics processing units or GPGPUs. Relative to traditional CPUs, these accelerators often provide significantly better performance and energy-efficiency for types of computation it is designed for. However, a domain-specific accelerator may not be suitable to run an entire application and require it to be accompanied by a general purpose CPUs – giving rise to heterogeneous computing. A key challenge to the successful adoption of heterogeneous computing is the difficulty in programming for heterogeneity considering vastly different programming models of typical accelerators like GPGPUs and general purpose CPUs.
In this talk, we will focus on GPGPU, a popular accelerator that is specialized in executing massively data-parallel programs. We will discuss key programmability enhancing feature on GPGPUs called shared virtual memory (SVM). We will show that SVM could help ease programming GPGPUs but can significantly slow down applications. We will then briefly discuss our idea of smarter scheduling of memory requests that could lead to 30% performance improvement. Finally, if time permits, we will conclude this talk by discussing some of the other relevant future research directions that are of interests to the speaker.
Biography : Arka(prava) is an assistant professor at the Department of Computer Science and Automation. His research interests lie in the hardware and software interface – specifically in computer architecture and operating systems. Prior to joining IISc, he was a researcher at the AMD Research at Austin, TX. Before that he earned his Ph.D. from the University of Wisconsin at Madison, USA.
Optimal Path Planning in Dynamic Environments
Abstract: Optimal planning of autonomous marine platforms is extremely important for their efficient use in a variety of scientific, security and humanitarian applications. In this context, we develop fundamental stochastic partial differential equations and efficient solution schemes for time-, energy-, risk-optimal paths of agents in strong, dynamic and uncertain currents. We apply our schemes and software to plan paths in a variety of idealized, realistic and real time operations of AUVs, gliders and ships in different coastal ocean regions. Notably, our planning rigorously incorporates environmental predictions and fundamental optimal control PDEs in a probabilistic framework.
Biography : Deepak Subramani is an Assistant Professor in the Dept. of Computational and Data Sciences at IISc, Bangalore. His research is at the intersection of computation, stochastics and geosciences, with a focus is to build holistic science-based data-driven computational solutions to complex engineering and environmental problems. Previously, he completed his postdoc, Ph.D. and MS from Massachusetts Institute of Technology, and Dual Degree (B.Tech+M.Tech) from IIT Madras. His recent research projects and publications can be found on his webpage: http://cds.iisc.ac.in/faculty/deepakns/
Data Markets for Smart Grids
Abstract: Conventional power grids are being transformed into smart grids with the help of computer technologies. This has led to the generation of huge amount of data. Management of enormous amount of smart grid data poses significant challenge to the data owners. Currently, generated smart grid data mostly stay with the owner of the data. Lack of a proper data exchange platform often prohibits the flow of data across various entities of smart grids. Exchange of data is necessary for efficient utilization of smart grid data. To address these issues, data market for smart grid has been proposed. This talk will give an overview of Data Markets for Smart Grids. Market structure, participants, commodities, attributes of data pricing, data licensing, data authenticity and other aspects of the proposed Smart Grid Data Markets will be discussed.
Biography : Sarasij Das received the Ph.D. degree from the University of Western Ontario, London, ON, Canada, in 2014. He was a Post-Doctoral Fellow with the University of Ontario Institute of Technology, Canada. He was with the Global Technology Centre, Schneider Electric India Pvt. Ltd., Bengaluru, India, and Power Research and Development Consultants Pvt. Ltd, Bengaluru. He was an Assistant Professor with the Indian Institute of Technology, Kharagpur, India. He is currently an Assistant Professor with the Electrical Engineering Department, Indian Institute of Science, Bengaluru.
Shayan Garani Srinivasa
Magnetic Data Storage: The Journey of a Bit
Abstract: Magnetic data storage has evolved from simple recording of a signal on a wire as demonstrated by Poulsen in the late 1890s to todays' recording technologies that aim at storage densities exceeding 4 Tb/Sq-inch. In this talk, I will cover the journey of a bit to be stored and retrieved from a magnetic storage medium. The talk will focus on various subsystems within this data recording technology, highlight alternative recording technologies from media/recording physics perspective, and the current research challenges therein. Towards the end, I will cover the basics of two-dimensional magnetic storage and its potential towards realizing high data storage densities driven by a systems approach.
Biography : Shayan Srinivasa Garani received his Ph.D from Georgia Tech, U.S.A, M. S from the Univ. of Florida, U.S.A, and the B. E degree from the Univ. of Mysore, India. Prior to joining IISc, where he leads a pack of PhDs and Masters' students, he was managing and directing the read channels research at Western Digital and Chairman of IDEMA-ASTC and co-chair of the overall technological committee. He is currently the Chairman of the IEEE Data Storage society. He is an inventor of several key patents that actually went into products, and was the lead architect for many generations of novel storage channel solutions for applications in hard disk drives (HDDs) and solid-state drives (SSDs).
From Simulation to Experimental Realization of Quadrupedal Walking in Stoch
Abstract: Stoch is a custom quadrupedal robot designed and developed in the Robert Bosch Center for Cyber Physical Systems at IISc. In contrast to existing approaches, we have used Deep Reinforcement Learning (D-RL) to realize walking in Stoch. In the talk, we will describe the methodologies used to address high dimensional statespace, transferability to real hardware, and hardware limitations. We first generated walking gaits in simulation by using policy gradient based approaches. Having obtained the gaits, we then realized walking in Stoch by using a low dimensional representation of these gaits, i.e., kinematic motion primitives. This type of methodology improves the transferability of these gaits to real hardware, lowers the computational overhead on-board, and also avoids multiple training iterations by generating a set of derived behaviors from a single learned gait.
Biography : Shishir is an INSPIRE Faculty fellow in the Robert Bosch Center for Cyber Physical Systems (RBCCPS) in IISc Bangalore. He received his Ph.D. degree in Mechanical Engineering (2016) from the Georgia Institute of Technology. His primary focus as a PhD student was on stability and control of walking robots. Shishir is currently interested in safety-critical control, stability of hybrid systems, and deep reinforcement learning for all kinds of robotic platforms.
|#||Poster Title||Name of the presenter|
|1||Battle of Bandits: Online Learning from Relative Subset-wise Preferences||Aadirupa Saha|
|2||Radial Point interpolation method using vector RBFs for Electromagnetic modeling||Aditya Sivaram|
|3||Auditory representation formation during language learning||Akshara Soman|
|4||Pitch-synchronous DCT features for speaker identification||Amit Meghanani|
|5||Unidirectional Single-Stage High-Frequency-Link Three-Phase Inverter For Grid Integration of Utility Scale Renewable Sources||Anirban Pal|
|6||Data driven propagation modeling for a class of IEEE 802.15.4 wireless devices in an indoor environment||Anitha Varghese|
|7||Low Power Analog Neural Network Framework with MIFGMOS.||Ankit Tripathi|
|8||HyDetect: A Hybrid CPU-GPU Algorithm for Community Detection ||Anwesha Bhowmik|
|9||Smart Meters for Protection of Modern Distribution Systems||Asha Radhakrishnan|
|10||Almost Unsupervised Learning for Dense Crowd Counting||Deepak B S|
|11||Hashing for Cross-Modal Retrieval||Devraj mandal|
|12||Linear Dynamical Systems with Sparsity Constraints: Theory and Algorithms||Geethu Joseph|
|13||Uncertainty modeling of electromagnetic systems with high stochastic dimenisionality||Gladwin Jos K T|
|14||High Throughput Asymmetric 3PC : A Secure Prediction Framework||Harsh Chaudhari|
|15||Impact of PLL on Harmonic Stability of Renewable Dominated Power System||Indla Rajitha Sai Priyamvada|
|16||Learning from Experience for Performance Optimization: Algorithms and Applications in Crowdsourcing, Cloud Computing and Enterprise Systems||Indu John|
|17||AdaDepth: Unsupervised Content Congruent Adaptation for Depth Estimation||Jogendra Nath Kundu|
|18||Soft Switched Multilevel Unidirectional High Frequency Link DC/AC Converter for Medium Voltage Grid Integration||Manmohan Mahapatra|
|19||Fast Actively Secure Five-Party Computation with Security beyond Abort||Megha Byali|
|20||Deep Learning for Improving Limited Data Photoacoustic Tomography||Navchetan Awasthi|
|21||Parallel hybrid smoothers in Multigrid Method for heterogeneous CPU-GPU environment||Neha Iyer|
|22||Neuromorphic In-Memory Computing Framework using Memtransistor Cross-bar based Support Vector Machines||Pratik Kumar|
|23||Unsupervised Representation Learning for Speech Recognition||Purvi Agarwal|
|24||Real Time Image Segmentation Using Neuromorphic Analog Pixel Array||Raja Sharma|
|25||Doubly-Fed Induction Generator based Wind Turbine Emulation||Ramu Nair|
|26||Group Delay Engineered Microwave Circuits for Analog Signal Processing Application||RiteshKumar|
|27||One-Shot Object Localization Using Learnt Visual Cues via Siamese Networks||Sagar Gubbi|
|28||Robust Stabilized Finite Element Scheme for Singularly Perturbed Differential Equations using Deep Learning||Sangeeta Yadav|
|29||Low Power Neuromorphic Analog System based on Sub-Threshold Current Mode Circuits||Sarthak Gupta|
|30||Deep Generative Learning-Based Compressive Sensing on Pre-Beamformed RF Data in Ultrasound Imaging||Saurabh Kumar Gupta|
|31||Language and Accent Recognition||Shreyas Ramoji|
|32||One-Shot Coordinated Control of Feeder Vehicles for Multi-Modal Transportation||Subhajit Goswami|
|33||Experimental Investigations on the Dependence of Switching Characteristics of IGBT Modules on Operating Parameters||Subhas Chandra Das|
|34||Dynamic State Of Health (SOH) estimation of Li-ion batteries using A.I. techniques||Surya Teja Tunuguntla|
|35||Hardware Emulation of a Long Transmission Line by High Frequency Power Electronic Converter for the study of Switching Transients||Sushmit Mazumdar|
|36||Distributed Processing Model for Temporal Graphs||Swapnil Gandhi|
|37||On the Exact Round Complexity of Best-of-Both-Worlds Multi Party Computation||Swati Singla|
|38||Switched Reluctance Machine for High-Speed Application||Syed Shahjahan Ahmad|
|39||Probabilistic forwarding of coded packets on networks||Vinay Kumar B R|
|40||Design and Analysis of Low Overhead Feedback Schemes for 4G/5G||Vineethkumar V|