AI FOR SOCIAL GOOD
“AI FOR SOCIAL GOOD”
by
Dr. Balaram Ravindran
Reconfigurable and Intelligent Systems Engineering (RISE) Group
Department of Computer Science and Engineering
Robert Bosch Centre for Data Science and Artificial Intelligence (RBC – DSAI)
Indian Institute of Technology Madras
Session on 16-06-2020 from 09:00 A.M to 10:30 A.M.
Session notes prepared by
Dr.G.Lakshmi Priya, M.E.,Ph.D.
Assistant Professor Senior Grade - I,
Dr. Henridass A, M.E., Ph.D
Assistant Professor (SG)
School of Electronics Engineering (SENSE),
VIT University, Chennai.
Home assistants, Alexa, Medical diagnosis, Super Humans, etc… many use AI. AI is everywhere. All advertisements are using AI which explores the users browsing history. Speaker mentioned about Deep Blue which was a chess-playing computer developed by IBM. AI was existing from 1950, where Alan Turing proposes a question on “Can Machines Think?”. Based on Turing’s question in 1956, The Dartmouth Summer Research Project on Artificial Intelligence was conducted. This was the first use of the term “Artificial Intelligence”. The Founding Fathers of AI was shown. AI has gone through several hype cycles
A.I. TIMELINE
Speaker discussed about the article released in 1958 and 2018 on “Machines can now see, walk and talk”. Given a set of images, machines can recognize the images and classify them as automobile, bird, cat, dogs, flowers, etc… But in recognizing these images, humans make some mistakes. Now, using different technologies, error percentage has been reduced.
Imitating the Brain
This is possible with so much of accuracy only when we feed more number of image data sets. Image should have more layers. MS ResNet had 152 layers to bring down the error rate to 3.6%. Artistic applications also discussed where in doodles are given and complete images have been drawn (Doodle to Painting, Style Transfer).
AlphaGo Master Chess games are based on AI. A computer program that defeated a champion Go player (Lee Sedol). AlphaZero: A general AI algorithm to play all board games. AI continues on Protein Folding - one of the hardest problem in biology – AlphaFold.
The current thinking is to succeed a large volume of data. AI has tremendous possibilities for Social Good. AI continues on several domains on Education, Healthcare, Transportation, Public Safety, Healthcare, Public Policy. To work on these areas Robert Bosch Data Science centre was established at IIT, Madras.
Education and Research at RBCDSAI IITM:
ARMMAN – IMPROVING THE EFFECTIVENESS OF MMITRA
ARMMAN is an India-based non-profit organization committed to improving the wellbeing of pregnant women, new mothers, infants and children in the first five years of their life.
mMitra is a free mobile voice call service by ARMMAN that sends timed and targeted preventive care information weekly/bi-weekly directly to the phones of the enrolled women through pregnancy and infancy in their chosen language and timeslot.
Salient Features
ARMMAN and BabyCenter has developed the messages that have been validated by The Federation of Obstetrics and Gynaecological Society of India (FOGSI) and National Neonatology Forum (NNF).
Frequency: 141 individualised voice messages of 60-120 seconds are sent with the following frequency
Pregnancy: Bi-weekly
First week after birth: Daily
Until Infancy month 3: Bi-weekly
Infancy month 4 to Infancy month 12: Weekly
Timed & Targeted: The information in the calls is timed to the stage of the pregnancy or age of the infant and is delivered directly to the women.
Chosen Timeslot: The calls are sent in the timeslot chosen by the women.
Preferred Language: Women can receive the calls in Hindi, Marathi, Kannada and Gujarati.
Repeat Calls: There are three tries for every voice message.
Missed Call System: If a woman misses all three calls, she can give a missed call to receive a call back.
Call-center: A trained counselor can be informed in case of a delivery, abortion or to change the phone number or timeslot.
Objectives:
Model Architectures and Training
DEVELOPING INDIA – SPECIFIC MODELS FOR FOETAL AGE ESTIMATION FROM GARBH – INI COHORT – HIMANSHU SINHA
Globally, preterm birth is a major public health problem. In India, 3.6 million of the 27 million babies born annually are preterm. Risk stratification of women based on multidimensional risk factors assessed during pregnancy is critical for prevention of preterm birth. An interdisciplinary Group for Advanced Research in BirtH outcomes- DBT India Initiative (GARBH-Ini) has initiated a cohort study of pregnant women has been initiated in May 2015 at the civil hospital in Gurugram, Haryana, India with the ultimate objective
to acquire deep fundamental knowledge on adverse pregnancy outcomes such as preterm birth (PTB) and fetal growth restriction (FGR) and
apply this knowledge to identify efficient and sustainable solutions that would assist in reducing associated mortality, immediate and long term morbidity.
Project KAATRU
Kaatru in Tamil stands for wind. According to the WHO, 9 out of 10 people worldwide breathe polluted air. Team Kaatru from IIT Madras aims to provide a solution for air pollution. We give our users real time data of the purity of air that they breathe. Similar to how a Google map provides a real time location of traffic in various roads, the Kaatru App will provide a real time spatial and temporal map of air pollution.
The project has four main components.
The first part is the setting up of the air pollution monitoring hardware device. This device is assembled by our team and is deployed at specific stationary points and on moving vehicles. This device comprises of various sensors such as the Particulate Matter content in ambient air, Temperature, Humidity, Air Speed, Multiple Gas Sensors etc.
These sensors are assembled along with GPS and GPRS systems which send the data about various parameters and ambient air to the central cloud server. There is also a memory card module to store the data locally.
Once this data from across various locations and time reaches the server, they need to be processed for us to make sense of it. The air quality parameters are then visualized by plotting it in a 3D map.
Data Analytics is used to derive insights from this huge chunk of data which are of huge business potential. Prediction algorithms and Time Series Analysis is used in order to fill any kind of missing data at any location at any point of time.
AI for Social Good
By Dr. Balaram Ravindran IIT-M by Dr. Henridas
Applications of AI
Google Maps
Zomato
Swiggy
Home Assistant like Alexa, Siri…
In medical Field
Radiology, Brain study, Monograms…
All advertisements in websites are based on Browsing history
AI is not New
Washington july 7 1958-- The navy revealed the embryo of an electronic computer today it expects will be
1950 Turing asks the question
I Propose to consider the question
Can machines think?
Alan Turing movie-The imitation game
1956 A new field is born
Dartmouth conference
Comments
Post a Comment