BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:CMS Seminars
SUMMARY: IBM research in Africa: an overview of the projec
 ts helping to build Africa’s future and career opp
 ortunities for Mathematicians - Dr Kamal Bhattacha
 rya\, Dr Osamuyimen Stewart and Dr Meenal Pore
DTSTART;TZID=Europe/London:20151020T170000
DTEND;TZID=Europe/London:20151020T180000
UID:TALK61930AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/61930
DESCRIPTION:Abstract:  Africa is poised to become a leading so
 urce of innovation in a variety of sectors\, with 
 an expected growth rate of 7% annually over the ne
 xt 20 years. IBM recognizes the huge potential imp
 act of research and smarter systems in helping to 
 build Africa’s future\, hence the lab is focused o
 n technology applications in a range of industries
  at the core of Africa’s growth. Employing some of
  the best scientists from around the world and is 
 partnering with universities around the world to d
 evelop and hire top talent.   There are career opp
 ortunities for Masters\, PhDs and Post Docs.  Dr K
 amal Bhattacharya (Director\, IBM Research – Afric
 a) and Dr Osamuyimen Stewart (Chief Scientist\, IB
 M Research – Africa) will give an introduction to 
 IBM Research – Africa\, and an overview of teams a
 nd projects at the Nairobi and Johannesburg.  The 
 research areas include:\n\nActive Learning\, : Acq
 uiring high-quality labelled data in resource-cons
 trained settings is difficult.  Active Learning ca
 n be used to drastically reduce the cost of gainin
 g insights from noisy background data.  For exampl
 e\, which household should be targeted in a health
 care survey based on the roof top composition of t
 he dwelling estimated from satellite imagery?\n\n 
 \n\nTransfer Learning: Africa is an incredibly div
 erse continent and even the most successful innova
 tion in one country may have very low performance 
 in other regions.  Transfer learning provides syst
 ematic approaches that allow researchers to re-use
  informative data streams for adaptation to a new 
 task\, new demographic\, or both.  For example\, b
 ased on a supervised classification model\, we pre
 dict that a particular cell phone user in Kenya wi
 ll be able to repay a micro-loan - will this model
  work in Nigeria as well?\n\nObject recognition: G
 iven the proliferation of affordable imaging techn
 ologies\, from drone aerial imaging to phone camer
 as\, how can such technologies be used to lower th
 e cost and improve the quality of evidence-based p
 olicy?  For example\, given the incredible rate of
  growth of Africa’s cities\, how can city planners
  use drone imagery to better understand changes in
  population density and socioeconomic status of co
 mmunities\, and hence forecast the demand for vari
 ous public services?\n\nDistributed Computing: Mob
 ile phones are the already-present incarnation of 
 IoT in Africa.  Harnessing this data requires new 
 approaches that can handle society-scale data.  IB
 M is globally investing in Apache® Spark™ to creat
 e advances in large scale data processing. 
LOCATION:CMS Meeting Room 4
CONTACT:Tina Jost
END:VEVENT
END:VCALENDAR