Top Banner
IIIT, Hyderabad: A Research University Rajeev Sangal, Director P. J. Narayanan, Dean (R&D) International Institute of Information Technology, Hyderabad Research Rendezvous March 24, 2007
28

IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

Oct 15, 2020

Download

Documents

dariahiddleston
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT, Hyderabad:A Research University

Rajeev Sangal, DirectorP. J. Narayanan, Dean (R&D)

International Institute of InformationTechnology, Hyderabad

Research RendezvousMarch 24, 2007

Page 2: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adAbout IIIT-H

• Setup in 1998, catalyzed by AP Government– Initial infrastructure from Government– Self-sufficient for everything else– A public institution; not private or government

• Vision: To produce the highest qualitytechnology and manpower.

• Research as the primary focus– At both UG and PG levels!

• Among top-10 tech-schools (Dataquest 2006)

Page 3: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adAcademic Programmes

• B.Tech Programmes: CSE & ECE– Research-led curriculum– Honours programme for research at UG level

• Research Programmes:– MS by Research: Minimal courses, high-class thesis– PhD: Some coursework, solid dissertation– In core IT areas and IT application areas

• Bioinformatics, Computational Linguistics, Structural Engg• Other PG programmes

– M.Tech: mostly industry oriented. Research is optional.

Exceptional Place for Exceptional Students!

Page 4: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adResearch Centres not Departments

• Research is the primary focus. Research centres canhave all resources needed for the work.– Faculty, MS/PhD students, UG students, Staff, etc.

• Centres designed to be mostly autonomous– Empowerment is the key

• Work on theory, algorithms, systems, applications• Performance:

– Publications, patents, systems, packages.– Convince the world!

Research that Makes a Difference!

Page 5: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adUGs in Research

• Excite UG students to take up research– Project orientation, early introduction to subject– Flexible curriculum– Honours programme to give taste of research– Dual-Degree B.Tech/MS for momentum

• Outcome: Appetite develops on tasting!– A third of the students take the research option– About 40 UG students joined MS/PhD

programmes of the institute in less than 5 years!

Page 6: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adUG Honours Programme

• 2 years of solid exposure to the stream area– Many electives in the area for background– Long projects for practical knowledge

• Mechanism– For good a`nd interested students– Join a stream at the end of fourth semester– Take a number of electives from the stream– Work 2 summers and 4 semesters doing research

• 2 semesters of the B.Tech project• 2 additional 4-credit projects as extra load

Page 7: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adOutcome of Experiments

• Critical mass of researchers in several areas– Overcoming handicaps of small groups

• Continuity of projects due to longer involvement– Honours programme keeps students for 2 years– Typical B.Tech/M.Tech projects do not overlap

• Break the chicken-and-egg problem– Need students for activity, need activity to bring funding

and students.– UG students bring activity and energy– Activity brings funding and research students

• High manpower impact on industry

Page 8: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adResearch Degree Programmes

• MS/PhD programmes started in 2002.Growing enrollment.

• Today’s numbers:– 33 PhD students (12 joined in 2006)– 100 MS by Research students (40 in 2006)– Over 30 MS graduates so far

• Vision for 2012:– Admit 100 PhD students per year– National production of PhDs: About 70

Page 9: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adEntry and Exit

• Make it easy to admit good students– Based on research record and recommendations– Work with a faculty and prove potential– 24x7 admissions, join anytime

• Set high exit requirements– Peer evaluation of research is a MUST– Let reviewers of top journals/confs do the work!– Easy committee evaluation with peer-acceptance

Page 10: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adCurrent Focus Areas

• Language Technologies: MT, IE, Speech• Visual Information Processing: IP, CV, Graphics• Data Engineering: Databases, Data Mining, Multiagents• VLSI & Embedded Systems, Compilers, Architecture• Security, Theory & Algorithms• Robotics• Communications• Software Engineering• Bioinformatics/Computational Natural Sciences• Structural Engineering & Building Sciences• IT for Education, Agriculture, Power

Page 11: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adArea-Wise Faculty & Students

100+331+10+32+16+42+06+0

18+111+45+6

32+317+10

15020031051520354515

3IT for X (Education, Agriculture, Power)111Software Engineering10

37Total

4Visual Information Processing28Language Technologies: MT, IE, Speech1

9876543

4Structural Engineering & Building Sciences3Bioinformatics3Communications2Robotics4Security, Theory & Algorithms4VLSI & Embedded Systems3Data Engineering

MS+PhDFaculty Hons

Page 12: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adSome Highlights

• DUC 2006:– IIIT’s summarization system was 1st in all categories

• RoboCup/RoboRescue:– Kshitij team placed 3rd in RoboRescue simulation 2005– Best new agent in 2006

• CSI-Nihilent eGovernance awards:– eSagu project overall winner for 2005-06.

• Hosting major academic events:– IJCNLP 2008, IJCAI 07, ACCV 2006, FSTTCS 2005,

COMAD 2005, ICON 2004, etc.

Page 13: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adResearch Philosophy

• Development of theory and algorithms– Publish paper, patents, etc– Be seen and recognized at the best forums

• Building research systems– Needs large UG/PG teams, continuity of activity

• Building field prototypes with industry– Academia has technology but can’t do alone– Needs team, research staff, mechanism

Page 14: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adCase: Information Extraction

• Language analysis for information extraction– $250K per CPU per year for commercial tools

• Language group at institute had the expertise– Demonstrated using existing toolkit– Made it deployment ready with trained students– Trained company employees on the toolkit– Supplied qualified students as employees

• Tool has a lot more applicability

Page 15: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adCase: BioSuite

• TCS developed a Bioinformatics suite underNMITLI scheme of CSIR

• Needed expertise in Graphics/Visualizaiton• Vision/Graphics group provided

– Students developed a prototype over summer– Participated in the design with the TCS team– Trained company employees in graphics– Supplied graphics trained UG students

• Domain expertise has many applications

Page 16: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adeSagu

• IT-based personalized agricultural extension system– Coordinators in villages visit each farm– Send digital images and text info to IIIT-H– Agricultural experts provide relevant advice– Advice delivered on farmer’s doorstep– Keeping track of progress

• Saving: About Rs 4000 per acre• 1000 cotton frams in 2004-05

5000 farms and 6 crops in 2005-06• Being replicated nationwide by MLA & IIIT-H

Page 17: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adSocial Impact

• RAVI: Screen reader for Indian languages• Virtual Laboratory

– Experiments with local content in local language– Tools to enable teachers design experiments

• Retinal Image Processing– Tracking drusen and retinal damages– Diagnosing diabetes from retinal scans– Working with doctors from LV Prasad hospital

Page 18: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adNational Impact

• Many solutions delivered to DRDO– Flythrough system for SASE– Image mosaicing and registration for ADE– Robotics and Vision with CAIR

• Information extraction for ADRIN, ISRO• Active role in MCIT’s Indian language consortia

– Machine translation, search, information extraction– OCRs, handwriting recongition

• Digital Library of India– 3 crore pages of interest to India made digital

Page 19: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adManpower Impact

• MS/Honours students at variousorganizations– GE JFWTC, MSR, Adobe, KritiKal– Nvidia, ATI, Riya, SET Labs, TCS ATC– Google, Yahoo!, MSR, Gridstone

• Our Vision:– 50% of high-end designers and developers in

product companies to be our MS/Hons students

Page 20: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adHow can the industry help?

• Believe in us– Our experiments/efforts are showing results– Plan assuming technology expertise and

qualified students are available• Support us

– Nurture your broad areas of interest at IIIT-H– Technical, financial, strategic, involvement– Look to us for technology, not just technologists

• Spread the message– Research is essential and pays off

Page 21: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adSupport: Traditional Methods

• MS/PhD fellowships to motivate students– Support for individual students– Impact is limited to the individual

• Research funding for activity– For an activity; has deliverables and schedules– Constrains theses and/or needs new people

• Faculty awards– Reward performance of faculty

Page 22: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adSupport: New Avenues

• Yearly Research Awards for Students– Some stipend, conference travel, books– We select and administer based on performance

• Area-based student fellowship– Support for 2 research students in, say, computer vision– We identify the students; s/he benefits

• Nurture an area– Funding for research in, say, grid-computing– Outcome: trained manpower, expertise, papers, prototypes.– Producing 5-6 expert students at IIIT-H costs less than recruiting and

training one person– If your area has research activity, you ultimately benefit

Page 23: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adConcrete Options

• Student Research Award (charity)– Rs 2 lakhs per award per year

• Area-based student fellowship– Rs 4.5 lakhs for an MS student support

• Nurturing an area– Rs 10 lakhs per year for impact of 1 faculty, 4-5

student (MS/Hons), thesis, and papers– Researchers look for employment, funder has

special claims to research

Page 24: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adSupport: More Options

• Travel support for faculty and students– Indian researchers missing from international forums– Visibility is important to students and faculty

• Access to advanced technologies early– To nullify the locational disadvantage of India

• Affiliate with research centres– Visit and interact frequently– Teach classes, guide students

• Hire UGs and send here for MS/PhD– Make them experts in theory and practice

• Tell UGs to do MS/PhD: You have influence

Page 25: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adHiring Research Students

• Hire them as research students– Give them respect as researchers– Call them over, listen to them, and hire them– We will circulate their folders

• Don’t treat them like UG students– Use a separate process and team to select them– Don’t follow aptitude test, programming test, etc– Instruct HR accordingly

Page 26: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adInvolving with Research Students

• Support them, involve with them during their study• Hire them after they complete their degree not in

the middle– Don’t kill the golden goose

• Give them jobs that utilize their skills/training– Treat them specially, you will benefit

• Support them to complete and continue theirresearch with IIIT-H– Be flexible so they can complete their thesis– Completing and presenting papers, extending their work

Page 27: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adSummary: What do we offer?

• Top-class technology and expertise in manyareas

• Well-trained research students– PhD: First one to come out in 2007

• Veritable researchers and experts in the area– MS by Research: About 30 have graduated

• Good training and expertise– B.Tech (Honours): 30-40 per year

• Exposure through many courses• Over 2 years of research/project experience

Page 28: IIIT, Hyderabadresearchweb.iiit.ac.in/~dnrnd/ResRendez07.pdf · IIIT, Hyderabad Outcome of Experiments •Critical mass of researchers in several areas –Overcoming handicaps of

IIIT

, Hyd

erab

adConclusions

• IIIT, Hyderabad is committed to research andhas a lot to offer to you

• Industry and IIIT-H can achieve a lot moretogether

• Several options for collaboration• Industry benefits through technologies,

expertise, and trained researchers