Postdoc Year at the University of California, Berkeley - PowerPoint PPT Presentation

1 / 17
About This Presentation
Title:

Postdoc Year at the University of California, Berkeley

Description:

... experiences of a post-doc year at the University of California, ... Digression to the year 2000... TEKES is looking for a ... 2000/month. Moved twice ... – PowerPoint PPT presentation

Number of Views:72
Avg rating:3.0/5.0
Slides: 18
Provided by: konstak
Category:

less

Transcript and Presenter's Notes

Title: Postdoc Year at the University of California, Berkeley


1
Post-doc Year at the University of California,
Berkeley
  • Konsta Koppinen, Ph.D.
  • Tampere University of Technology
  • konsta.koppinen_at_tut.fi

2
Overview
  • Personal experiences of a post-doc year at the
    University of California, Berkeley
  • Chronological order

3
Personal Background
  • Born 1971
  • M.Sc. University of Tampere, mathematics 1997
  • Worked at the Institute of Signal Processing,
    Tampere University of Technology since 1995
  • Accepted to TISE graduate school in 1997
  • Work on
  • Managing various signal processing and speech
    processing projects (TEKES, Nokia, EU)
  • Lecturing of speech processing/coding/recognition
  • Research on polynomial predictive filtering
    (Ph.D. topic)
  • Dont do this kicked out of TISE in 2001
  • Time spent on projects/teaching
  • Funding from projects

4
More Background
  • Wrote Ph.D. manuscript (monograph) in 2002
  • Dont do this 2 no journal paper in Ph.D.
    manuscript
  • 2nd reviewer needs improvement
  • Spring/summer 2003
  • Regathering of motivation for writing the Ph.D.
    thesis
  • Organization of a big speech processing TEKES-
    project (PUMS) after the end of another one
    (USIX)
  • TEKES National Technology agency of Finland, a
    big funding agency for technological research
  • Writing of a journal paper
  • Big revision of Ph.D. manuscript

5
Actually Relevant Background
  • 7.3.2003 video conference presentation of speech
    research at TUT for prof. Nelson Morgan of ICSI
  • Previous collaboration TUT/ICSI
  • Asked around TUT for interested people
  • hmm, I could go myself
  • Digression to the year 2000...
  • TEKES is looking for a university partner
  • Several international universities are considered
  • Berkeley/ICSI is chosen
  • silicon valley close by
  • looots of venture capital
  • high-quality research
  • 4 places reserved at ICSI for Finnish researchers
  • Projects are encouraged by TEKES
  • Project on noise robust multilingual speech
    recognition with TUT started

6
Internal debate
  • No projects,no teaching, just research for a
    year
  • California
  • Good opportunity
  • I could try to include it in the PUMS-project
  • I have to write my Ph.D. first
  • If I dont go now, Ill never go
  • OK, Im going!

7
Planning, spring/summer 2003
  • Discussions with TEKES
  • Need
  • Research plan
  • Funding
  • A go-ahead from UCB
  • Making a research plan
  • I had a clear research goal in mind
  • Training of neural networks for speech
    recognition using evolutionary algorithms
  • Discussion with prof. Morgan during a visit
  • Discussions with the speech project consortium
  • Funding as a part of the project OK
  • Lots of work on finishing Ph.D.
  • Arranging for a visa through ICSI
  • Approval from UCB -gt embassy
  • J-1 researcher visa, no salary from USA
  • Employed by TUT the whole time
  • 2 year rule cannot apply for a work visa in the
    USA for 2 years
  • Housing arranged through ICSI
  • 2000/month

8
Final steps
  • Ph.D. thesis defense 4.10.2003
  • Put stuff in storage
  • Flight to San Francisco on 16.10.2003
  • Housing arranged beforehand
  • Office already set up at ICSI (computer phone)

9
What is ICSI?
  • Non-profit research organization affiliated with
    (but not part of) the University of California
    Berkeley (UCB)
  • Several international visitors
  • Headed by prof. Nelson Morgan
  • Staff about 90 people
  • 4 research groups
  • Speech research (mostly recognition, prof. Nelson
    Morgan)
  • Networking (Internet-related, prof. Scott
    Shenker)
  • AI (mostly natural language, prof. Srini
    Narayanan)
  • Algorithms (computational complexity, prof.
    Richard Karp)
  • TEKES has 4 paid positions for visiting Finnish
    researchers at ICSI

10
At ICSI
  • Talk with prof. Morgan and researchers
  • About 28 people on the speech team
  • Got to know the relevant background work
  • Training of neural networks for speech
    recognition
  • Novel feature vectors used in speech recognition
  • Speech databases
  • Lots of neural networks used at ICSI for speech
    recognition
  • My topic training of neural networks for speech
    recognition using evolutionary algorithms
  • Even so there was quite little overlap with ICSI
  • I was offered to be involved with ICSIs projects
  • Preferred to work on my own topic
  • Was somewhat isolated
  • Which I didnt mind
  • Speech lunch every week
  • Lunch a presentation
  • Also videoconferencing with collaborators

11
(Personal) Observations on Work
  • Very useful to see how another research institute
    works
  • Organization, people, projects, funding,
    infrastructure
  • Actual research is similar
  • Reading, coding, feedback, writing papers
  • Researchers seemed to interact more than in
    Finland
  • People talk a lot more
  • Discussion seemed very open
  • Perhaps a little less independent
  • Often B.Sc. and M.Sc. are done at different
    universities
  • The research depends on good individuals
  • Research seemed very directed by the funding
  • In particular, to meet specific improvements for
    speech recognition
  • This discouraged novel ideas
  • Resulting contacts are useful
  • Plus obviously the technical knowledge

12
Non-work stuff
  • It helps a lot to have some friends right away
  • A couple from Finland came over at the same time
  • Also an active group of local Finns
  • Good for practical stuff where to buy things,
    how to get around, borrow a car/tools/etc.
  • Also good for loneliness
  • Easy to talk to people
  • But hard to get close to them
  • Good idea to have some things beside work
  • Can be hard on relationships
  • Culture shock was worse than I expected
  • Took about 6 months to feel at home
  • Even after that IQ feels 20 points less than
    usual
  • Diffusing/getting used to another culture is a
    slow process
  • Everything has a definite end-date
  • Ill call you I wont
  • A different set of behind-the-scenes expectations
    and habits
  • Useful things TV, phone (landline), computer,
    DSL, car
  • Stay more than a year car is tax-free
  • Big pastime of visiting researchers discussion
    of what car to buy

13
  • Slkfdj

14
(No Transcript)
15
California is a beautiful place...
16
Summary
  • Very useful year!

17
Links
  • TEKES http//www.tekes.fi
  • TEKES/Berkeley http//www.tekes.fi/berkeley
  • ICSI http//www.icsi.berkeley.edu
  • Embassy http//www.usembassy.fi/
  • SF Bay area housing/stuff http//www.craigslist.o
    rg
Write a Comment
User Comments (0)
About PowerShow.com