@ IIT Madras

The AI4Bharat Initiative

Associate Professor, IIT Madras
PhD, IIT Bombay

 

Mitesh M. Khapra

Pratyush Kumar

Anoop Kunchukuttan

Researcher, Microsoft Research
Adjunct Professor, IIT Madras
PhD, ETH Zürich
 

Researcher, Microsoft
PhD, IIT Bombay

 

100+ academic papers

30+ US patents

Recognized by

Experience

Team

Research expertise

Academic, industry, and startup experience

Mission statement

Bring parity with English 
in AI tech for Indian languages 
with open source contributions

We want to be the Apache for Indian languages AI stack

मी तुम्हाला प्रतिकिलो 100 रुपये देऊ शकतो

100 கிலோ அரிசியை விற்க விரும்புகிறேன்

What would it take for a farmer in TN to talk to a wholesaler in Maharashtra?

We need to solve four fundamental problems

Speech Recognition

Language Understanding

Machine Translation

Speech Synthesis

Why is this problem hard?

Open source all datasets and AI models with permissible licenses

Build a coalition of partners to collect datasets and deploy models

বা

हि

తె

कॉ

ગુ

ने

कों

सं

ਪੰ

सिं

اُر

मै

Scale and Diversity

Unique language phenomenon

mujhe bahut confusion hai

Scarcity of resources

Lack of basic speech and NLP tools

Named Entity Recognition

Sentiment Analysis

Topic Classification

Content Filters

Keyboards

Spell checkers

Indian languages lag on academic benchmarks

Speech Recognition

Language Understanding

Machine Translation

Speech Synthesis

Why this is foundational for the social sector?

Open source all datasets and AI models with permissible licenses

Build a coalition of partners to collect datasets and deploy models

Indian language + voice support =
Key to interface Bharat

Why this unlocks commercial value?

Open source all datasets and AI models with permissible licenses

Build a coalition of partners to collect datasets and deploy models

Creating solutions based on local needs and behaviour is critical to improving user engagement

 

Consumer Survey,
Feb 2018, Bain Analysis

Voice could play a
pivotal role in enabling
e-governance and
bringing next 300 million
Indians to digital
platform

Nasscom Survey,

2019

 Indian language users are expected to account for 75% of  India' internet user base

KPMG/Google Survey,

2018

Chat & entertainment

Social media & news

Digital write-ups, payments, e-governance, e-commerce, search

Have we solved parts of this problem?

Speech Recognition

Language Understanding

Machine Translation

Speech Synthesis

Mined 33 million
parallel sentences

Built billion parameter translation models

Our models for En to 11 Indian languages beat
all models (including
Google, Microsoft)

Who is using our translation models?

NGO for book translation

Govt. for website translation

Fiction book translation

Judgment
translation

Feedback from users

 

“The quality of translations is significantly improved. I would say this is more so for the legal document where there were complicated sentences/cards which were translated very well. The syntax mostly did not falter even in the face of multiple ideas/information contained in one sentence.”
 

“The amount of time spent on correcting/improving the translation has dropped.”
 

“THIS IS VERY PROMISING. AMAZED BY THE SPEED.”

 

“I TOOK A PRINTOUT AND WENT THROUGH EVERY LINE. THE TRANSLATION IS 98% ACCURATE AND HIGHLY SATISFIED.”

What we want to do in the next 5 years?

Language Understanding

Machine Translation

Speech Synthesis

Speech Recognition

Create and release datasets

Train and open source AI models

Build deployable tools

Outreach to
drive use

Driving Science

Driving Adoption

What are our milestones?

Year 1

 Release models and datasets for MT and Language Understanding 

Release speech-to-speech translation systems

Release models and datasets for Automatic Speech Recognition

Year 2

Year 5

Set up "The AI4Bharat Initiative"  @

What do we need in the next 5 years?

Data

Team

Infrastructure

Outreach

Generate rural employment

Train 50 researchers (Residents, MS, PhD)

Bring compute parity to IITM

Handhold startups to innovate

Cost Breakdown

1 Million parallel sentences in 10 languages @ INR 20 per sentence

5000 hours of transcribed speech data in 10 languages @ INR 2000 per hour

NLU benchmarks with  a million sentences annotated in 10 languages @INR10 per sentence

20 Cr

10 Cr

10 Cr

Data Collection activity will be equally spaced over the next 5 years

INR

Cost Breakdown

CTO (PhD + 3-5 years experience)

3 ML Researchers (PhD)

8 AI Residents  (B.Tech)

60 L

72 L

48 L

~4 Cr INR per year for the next 5 years

36 L

4 ML Engineers (B.Tech + 2-5 years experience)

3 Principal Investigators

120 L

COO

24 L

5 Admin Staff

12 L

Chief Evangelist Officer

24 L

Annual Salary (INR)

40 Cr

Cost Breakdown

6 DGX A100 Servers

9 Cr

Office Infrastructure (Space/Desktops/Laptops/Printers etc)

Cloud infrastructure (for storage, hosting services)

3 Cr

3 Cr

20 Cr

40 Cr

INR

Cost Breakdown

Conduct workshops (2 per year)

2 Cr

20 Cr

40 Cr

14 Cr

Startup Enablement Program (1 per year)

AI4Bharat Grand Challenges (60L per year)

5 Cr

3 Cr

INR

The AI4Bharat Initiative

By One Fourth Labs

The AI4Bharat Initiative

The AI4Bharat Initiative

  • 592