Archive for the ‘Research’ Category

Early Spring Bloominghack!

Tuesday, March 23rd, 2010

- Alex is porting the rhyme-scoring code from the poetrybot to Python
- but right now, it only does end rhymes, which isn’t particularly interesting. (but it’d be pretty easy to expand to internal rhymes…)
- in research news, he plugged toulbar2 into the dependency parser (which uses constraint solving to do a parse — and now pretty soon will do “soft constraints”, or just “preferences”) — and is now thinking about how to find out what the weights on the constraints should be, with machine learning.

- Lindsey read about information flow security in a PL context (so, like: languages that support security levels in the type system. eg: jif)
- also: went running
- research news: working on translating a language with dependent types to continuation-passing style. Also, thinking about how to formally describe (like, with automated theorem-proving) the interactions between static and dynamic languages.
- “I’m excited that the papers that I’m reading to get ideas from were published in 2010.”

Hackfest 10000110

Tuesday, January 13th, 2009

Nirmal (Patel) is here!
Telling us about his research and trips to Korea for a project.
He’s been reading for quals.

Alex is hacking Haskell, doing the Accordian problem.
Doing IO.
Working on chapter 4 of Real World Haskell.

Rob is scheduling a Martial Arts Movie Party.
He uploaded holiday photos. ( New Year’s Eve and beyond )

Mark is talking with Nirmal.
He worked on Corkbird’s Processing/Java -> JSON serialization over the weekend.

Lindsey made a page for her academic exploits, started classes (incl. a compilers class), and is now an Associate Instructor (aka TA) for the undergrad programming languages class.
Also: working on Project Euler, just finished #4.

Hackfest 1101110

Wednesday, March 5th, 2008

Martin:
– hacked on D and GTK-D
– making a texteditor (textmate+vim clone)
Erik:
– entering exploratory phase of magic web framework
– tentatively titled “stellar”
– derived attributes, caching, everything in the model
– a big “F.U.” to M.V.C.
Mark:
– wrote a stack-based language: “yonth”
– adding more primitive operations tonight: greater_than, not_equals
– reading about Joy language, quoting
Alex:
– scrambled for launch
– successful except for minor communications snafu
– so it became an internal launch
– reading sicp, and mythical man month
Lindsey:
– received shiny new macbookpro at work
– received shiny emails from future professors
– ate quinoa
Anna Marie:
– wetting her pinky toe in web design
– working on a portfolio site for a friend
– figured out text-align: justify in Dreamweaver
– (still her means of coding)
Drew:
– released documentation for Miru: http://miru.enterthefoo.com/

Hackfest 1100111

Wednesday, January 16th, 2008

    Martin:
        contract with blueheat,
        their development process is scary,
        wrote python code to solve a scheduling problem
    Alex:
        showing off gwt google code wiki, now non-secret,
        is going to Portland this weekend
    Graham:
        refactored Mused such that it should be easier to add new features
        (not software features, but audo features),
        wants to submit a paper to ICMC, needs a good idea
    Erik:
        forecaster.ws, looking at php weather,
        has converted to distributed version control!! (git)
    Mark:
        used craigslist to find housemates, all were human,
        writing haskell to play slitherlink
    Sonali:
        maintaining choices quilts code,
        JDBC, System.out.println
    Tejus:
        doing the home-owner’s association thing,
        trying to avoid becoming corrupt
    Kelly:
        ga state is ok,
        first day was rough, trouble getting back home,
        no marta money, phone dead, had erik’s keys

Hackfest 1100101

Tuesday, December 18th, 2007

Since last week…
Erik: "Nothing! On purpose!" … Erik is largely On Vacation.
Alex: Been working on the documentation browser thing for code-google-com, which is no longer secret! He’ll make everybody take a     look at it when it launches, which is Soon. Also been working on the scrabble bot. And he got some cool anagram code from ZachG. Also, paper based on his masters work got into CHI.
Humza: Reading the new Lawrence Lessig book.

Tonight…
Alex: Looking for good information about event loops and callbacks in JavaScript, particularly on IE, where there’s occasional reentrancy weirdness.
Erik: Relaxing.
Humza: www.freerice.com, talking about Arabic linguistics — and culture and language in general, Sapir-Whorf.
Wadner: (the r is silent) … spoke with Humza about culture and identity, online and off.

(where is everybody?)

Hackfest 1011011

Wednesday, October 10th, 2007

New Member:
    Nate:
        working on startup: Wamily
        make-your-own social-network with widgets
        targeted at young people, customizable
        Emory graduate, wants to move to the valley
        runs Ubuntu, uses RadRails

Since last Time:
    Erik:
        working on pipaya
    Kelly:
        read a lot of books:
        "lilith’s brood" – about alien sex, a trilogy
        "you shall know our velocity"
        "children of men", "solaris",
        wrote one of the most powerful introductions to a paper that she’s ever written
    Drew:
        painting a mural of a mountain lion, a rabbit, and a pig,
        in the baby’s eventual room,
        working on gnarly legacy projects,
        things that were once considered a good idea
    Mark:
        moved into a house in decatur,
        processing sketch of fuzzy ellipse ray-tracing,
        read about juno-2 constraint system
    Alex:
        watched another machine learning lecture,
        reading about linear algebra,
        ada-boost, support vector machines, bagging
    Zach:
        BEST robot,
        designing spellchecking algorithms
    Joe:
        week of Tex-Mex, tonight Taco Bell
    Rob:
        configuring his keytar,
        put knobs for the keytar
    Gregg:
        making and repairing lamps in his house,
        found on the side of the road, rusted out,
        spent three hours searching for a lamp shade
    Stuart:
        finished design,
        will be sent to the photolithography place,
        will start production in a few weeks
    Tejus:
        etl (extract-transform-load), pictures

Tonight:
    Erik:
        working/talking with Nate
    Kelly:
        reading "you shall know our velocity"
    Drew:
        working on his legacy projects,
        feeling guilty about not working on oilrig
    Nate:
        working/talking with Erik
    Mark:
        preparing bar camp presentation,
        feeling guilty about not working on oilrig,
        taking lots of notes
    Alex:
        teaching about machine learning,
        "the kernel trick"
    Zach:
        learning about machine learning
    Joe:
        cranking voter files,
        reading about postgress internals,
        partial indicies
    Rob:
        likes being able to move around while music-making
    Gregg:
        sketching
    Jim:
        reading film textbook
    Stuart:
        "you can exert some physicality" – to rob about keytar,
        explaining photolith materials to mark,
        work on a song, drinking tea

Hackfest 1011010 Mortem

Wednesday, October 3rd, 2007

Since Last Time:
    Erik:
        releasing an app that converts a rhapsody-listening-feed to last.fm,
        rhobbler.com
    Mark:
        books: "Japanese for Busy People 2",
        "Domain-Driven Design", "Beautiful Code"
        learned some with scala (jvm-based functional language)
    Alex:
        triangular wave weekend,
        watching machine learning tutorial online,
        plsa = probabilistic latent semantic analysis,
    John:
        is building an outliner,
        may use google gears,
        will return with gwt questions for alex
    Rob:
        played a show at parkgrounds
    Stuart:
        doing research,
        working on the new album,
        due by the end of the month

Tonight:
    Erik:
        the server is good now
    Mark:
        reading papers about haskell,
        writing a soft constraint solver in nodebox
    Alex:
        has a stack of papers from acm portal,
        wants to learn statistics (and thus R),
        wants to learn linear algebra (and thus Matlab)
    John:
        discussing ajax,gwt,xaml,silverlight,flash
        no internet, so he goes home
    Rob:
        preparing for laptop battle,
        going to learn reason tonight,
        it’s what stuart uses, and rob likes the cheese
    Will:
        explaining how CAD programs do geometric constraints,
        degrees of freedom
    Stuart:
        hanging out

Hackfest 1010100 Postmortem

Wednesday, August 22nd, 2007

21 August 2007 – billings talks about tries, datastructure for autocomplete

new member Hamza! from Google ops
also we discussed the differences between art, science, engineering

Since Last Week-
Mark- is off hiking the Appalachian trail with his brother
Alex- talking with little zach about data structures
 -catalogued differences between java 1.4 and 1.5 by diffing javadocs
Kelly- started classes at UGA, human sexuality, speculative fiction class
Erik- working on projects that collects server statistics, pipaya
 -talking with Gregg about a manufacturing resource planning for his lab
Gregg- is now a grad student. might have a boyfriend.
 -fillings orders for sensors in the lab, furnace maintenence, training new guy
Billings- made progress on his amp. designed a new bias provider,
 -tested in Spice, ordered transformer
Rob- work stuff, new songwriting
Stuart- did some new project! Alex will tell me about it
Graham- no progress on getting visa faster. started packing up. investigated rooms.
Hamza- working on model to judge effectiveness of machine health model

This Week-
Alex- trying to get fluxus set up again
Kelly- is reading Solaris, reading On the Road by Cormac McCarthy
Erik- working on pipaya
Gregg- putting out feelers for database system to track parts in the lab
Billings- reading stories from PG Wodehouse
Rob- writing tanooki zoo songs
Stuart- doing some exciting project
Graham- must get Mused ready for presentation at ICMC by XRae

Hackfest 1000110 Mortem

Tuesday, May 1st, 2007

01 May 2007 – atlhack, your connection machine

met AJ, grading papers – grading papers for intro to MBA at UPhoenix. "Born to Party" t-shirts

Sonali means like Gold! Tejus means bright like the sun!
Graham’s name means "gravel area" or "grey homestead".
Mark means a dot on a plot or visualization. Will means great protector.

Since Last Week-
Will- job search stuff, met with Coke people. debugged BASIC for running beverage dispenser, with a Y2K bug, doesn’t deal with the midnight wrap. now it uses millisec since 1901.
-finished with reading Jason’s entrepreneuring books.
Mark- made bricks in nodebox. wants to make cornered bricks.
-doing visualization for high dimensional dataset.
-was chased by a Jeep and slammed his finger in a door.
-met Jay Jackson, a Qaboomer, starting brewpot.com
Tejus and Sonali- got highlights in their hair.
Graham- got dynamic queries working on the max pitch class profile.

Mark likes Mira Nair imdb wp documentaries and dramas, movies about Bombay street kids.

Graham
-loves Mark Guzdial’s blog on CS education- GOTO considered useful for learning.
John Maeda – artist at media lab – advisor to Processing. He used to write programs without loops.
Functional languages do not perhaps support the tracing-through metaphor of languages.

Tonight-
Will- is reading the Connection Machine, after having a parallel processing discussion with Mark.
Mark- working on building walls and brick houses.
Tejus- wants to learn how to use Rake (ruby build tool) more effectively.
Sonali- Mark will help Sonali prepare for an interview which requires three years of javascript.
Graham- finishing up his first draft of the paper.
Emily- is really busy and doing finals.

Senior Seminar / Final Quarter Chapter 1

Friday, January 19th, 2007

Hello everyone. I’m Mike. I’m studying mathematics at University of California at Santa Cruz. This is my last quarter and I have a talk that I have to give, and a senior project that I have to do.  The idea of these posts will be to track my progress on this project, and my progress in my other classes. I hope that this will help motivate me to do my very best in my last year in university.

Here is the class rundown in order of difficulty:
   Senior Seminar
   Real Analysis
   Latin 2
 
Here is the broad view:
   Senior Seminar:
       Give a 35 minute talk on a mathematical topic, on February 7th
       Turn in a 10 page paper on the topic by the end of the quarter.

    Real Analysis:  
       Weekly homework. Due Wednesdays.
       Reading.
       Tests.

    Latin 2:
       Homework for class 3 times a week.
       Reading.
       Tests.

Listing these is already starting to make things feel more manageable.

Now for some more details on my Senior Seminar:
    I want to study polyphonic source separation algorithms. Through what little research I have done so far I have realized that this is too broad a topic to fit into a 35 minute talk. So, my first task is to narrow down what I will talk about. As it stands right now, maybe a short talk on the Discrete Fourier Transform would be good. Right now, the basic structure might go something like this:

1. Explain in broad terms how digital audio works.  Samples, sampling frequency, possibly the sampling theorem, Nyquist frequency, etc.
2. Explain the mathematics of the Discrete Fourier Transform. Maybe, talk about the Fourier Transforms uses in other branches of math and science.
3. Explain what polyphonic source separation is, and how the Fourier Transform is used as a tool in polyphonic source separation algorithms.

    Somewhere in there I want to talk about psycoacoustics, and cognitive science, but maybe there is not enough time to do that and I will have to save it for the paper.

In conclusion I will leave a list of things that I need to do in the following week:
    Senior Seminar:
       Continue reading Tuomas Virtanen‘s thesis on polyphonic source separation.
       Write a possible outline, and bring it to my professor on Monday.

    Real Analysis:
       Complete homework assignment. Do at least 1 problem every day.
       Finish reading chapter 1. Begin Chapter 2.

    Latin:
       Catch up on backlogged homework. (Tomorrow!)
       Stay caught up. Do a little every day.

Next time I will review my accomplishments, and I will describe a technique developed by my parents to help get ideas down on paper.

-Mike