Looking at the images with misclassification results, there is no obvious reason to explain the errors. Each image (60 x 60 pixels x 3 channels = bunch of numbers) is transformed in a feature vector with 8 numbers. When plotting these features for each class of animal, we found out that only 3 of them are enough to distinguish among our current classes. This is true for both grayscale and true color images. So we reduced our feature dimensionality to 3 and got nearly the same results. On the classification process, first we train the algorithms with known images, and then test with a different set of images. To increase the size of the test set and still have enough images for training, we'll randomly select a part of the whole data set for test and do the training step using the other part. This will be done a lot of times and then we'll have a better statistical study of the classifier. I'm sorry for writing so many technical stuffs but, as the project makes progress, it's difficult not to do that. Apart from the project, going kayaking was very cool, and I'm looking forward for Salinas rodeo this weekend.

This week I got a nice break from lab work.  I worked on some reading about real-time PCR and even more articles for my paper.  I also compiled a bunch of 16S sequences and ran an alignment using ClustalW after struggling with the software for awhile.  The data from this will be extremely helpful when I'm trying to design primers for my qPCR that are specific to my mystery bug.  I was out sick yesterday, but still was able to get in a bit of reading.  Next week, I will finally be sending out my ligation mixture for sequencing.  Also, I will be transforming and E. coli cells with my plasmid.  I hope to also get my qPCR primers ordered.

This week I have been organizing all the CTD data in an excell sheet from the videos of the eggcases. I am mainly working on constructing good, understandable graphs even though excell does not do excatly produce the type of graphs I want. Hopefully excell will cooperate soon. Also, I am getting prepared to run DNA and PCR test on some tomopterid samples. The idea is to match the larval samples with some adult samples in hopes of having the same species of tomopterids since there are several species known. Anybody who would like to give advice about PCR's please please let me know =)

Also, I really enjoyed the kayak trip and seeing all the beautiful scenery. It is a shame to have a power plant so close to a reserve and all the natural inhabitants (cute sea lions and sea otters), but overall it was a wonderful break from the ordinary intern meetings.

This week I've begun my stint of meetings with potential SF Bay partners. Thursday (today) I'm meeting with USGS folks in Santa Cruz and with the Dredge Manager at the BCDC (Bay Conservation and Development Commission) in SF to talk about current profile data around Angel Island. Monday/Tues/Weds this week I spent preparing for these meetings and reading background info on West Coast Ocean Data Needs Assessments. I'm working to push forward some of my potential projects -- meeting with folks, sending them data to look at, brainstorming products. Hopefully a good project will come out of it.

On Sunday I'm going up to Tiburon (in Marin County, just north of SF) to work on some instruments that we're deploying around Angel Island (in SF Bay). I'll be there Monday and Tuesday. And get to meet with the NERR (National Estuarine Research Reserve) Research Coordinator, to discuss data needs/sharing.

So, I won't be there for Monday's meeting! (I just realized that...) But, I think this Tiburon meeting had the potential to be very valuable for my project. See you all Wednesday!

Hey Megan - this is George - sorry to have missed ya on Monday. I have a present for you from MBARI...

This week I switched gears and focused on COVE. I managed to get a grasp of the ''undocumented'' python scripts written by Keith Grochow (the man who developed COVE). I tweaked the scripts so that I can show the temperature data from July's front following AUV mission in a geospatial context. After having to go through the trouble of transforming the data in a 'netcdf' format I was able to show a surface like data on COVE. Only one problem, COVE kind of freezes. For some reason I haven't figured out COVE doesn't respond at almost all...I have emailed the situation to Keith and I am waiting for his response. 

On the other hand, I also did some serious Matlab scripting to generate the required files for Google Earth and I managed to show a surface like temperature profile in it. I also wrote a script to show the different behaviors performed by the AUV at different stages during a mission visually.

Next week what I will be doing is packaging all the things that I have written so far in a neat and easy to use interface and see what I can do with COVE. Once I finish all this, I will start looking deeper into TREX to come up with a way of interfacing this. 

Extremely frustrating week.

First, sat down and mapped out the rest of the Ysi instrument driver for SIAM. The 6600 driver will definitely need some work but I think I can extend most of my previously written classes. This part of the project should be short and sweet when I can find the time to sit down and code.

However, the majority of my time has been debugging the PC-104 board. After a meeting with my mentor and his associates it was decided that the kernel upgrade (2.6.21) was much more beneficial than trying to fix JamVM. The plus side is being able to use the Sun Microsystem embedded JVM. Cross-compiling the kernel was challenging but doable. The hard part now is to have the kernel recognize my root file system and correctly operate. This, of course, is not the case. As this is hardware specific I've poured through the PC-104 manufacturer's woefully limited documentation and have had support over the phone. Definitely not a fan of their treatment so far and still waiting for an e-mail about my latest problem...

I'm between a rock and a hard place right now.

Edit: I just got it! BOUYA!! 

This week I've just been trying to optimize the PCR reactions on my machine. I think we've figured everything out, so next week I might be able to do some DNA extraction using the ESP and run PCRs of those environmental samples. Today I filtered a bunch of seawater samples from Santa Cruz and got to look at all the little critters under a microscope (very cool). Todd hooked me up with this awesome set-up so I was able to take pictures and videos of the cute little guys. It was pretty sweet.

This week had its ups and downs.  It turned out that I wasted about 2k sequencing my primers (and a good chunk of my vector)-- so I had to do a gel purification of my PCR for sequencing.  After doing the gel purification etc. I ligated my DNA... however, when I ran a PCR of my gel-purified product, I still had excess bands.  Some of us think that this is a just an artifact of PCRing a PCR product (from gel purification), but just to be sure that we have the correct insert we are going to run the ligated product through bioanalyzer. 

 On an upside, however, ARB is up and running.  Totally exciting, as I was fairly certain for a while that it was never going to run on any computer that I touched.  But, Adam and Chris Preston came to my rescue-- and now it is running beautifully.  I have already uploaded and manually aligned a whole bunch of my sequences.

Next week Eunsoo is coming to give a lecture, and Alex has scheduled some time for me to work with her, as we are working on the same project.  

This week was pretty good and very busy. I finished a lot of the machining I had to do for my project and I was even able to run some tests with the new parts later in the week. I was glad to get the paper outline just so I could start mapping it out. there is still a lot of work left for me to finish but I think it can be done.
I have also finished some of my pelton prototypes, hopefully I will be able to send them out next week so they can be tested ASAP. I think there should be a pretty noticeable improvement in efficiency with the new wheel compared to my current setup. The initial ruby orifice tests also seemed promising. I was able to get more power out of the wheel using the ruby orifice instead of the standard steel orifice. Further tests I hope will validate these initial results.
Next week I will be working on my paper, running more orifice and wheel tests and doing some more reading. I'm going to try and look more at the math that deals with the specific problems I've discovered in my tests. There will most likely be some thing unforeseen that comes up with modeling, so I have budgeted a little extra time for that.

This week, I got most of the "mark new sightings" page on Jellywatch up and running!  I wrote (and by "wrote" I mostly mean "copy/pasted from google map api documentation and altered as necessary") a new thing in javascript that allows a user to search a map by typing in an address.  The address is sent to google for parsing and then google sends back a lat/lng coordinate for the address.  The user can then drag the marker around so that he/she can pinpoint exactly where the jellies were sighted.

My goal was to finish up the "mark new sightings" page for this week, and I've *almost* finished it, but I hit a little snag along the way.  We want users to be able to upload images onto jellywatch, but that requires the GD library, a library for php that allows for image manipulation and thumbnail creation.  Because I do not have root access on the webserver where we are hosting jellywatch, I asked IS to install the GD library, thinking that would fix things.  Unfortunately it did not-- I now need IS to re-configure our PHP installation such that PHP sees that the GD library is installed.  I'm sure this sounds like a bunch of gobbledegook--it was gobbledegook to me just two weeks ago! 

Also two weeks ago I didn't know javascript but with the help of the google maps API's well-written example code I've been able to figure stuff out reasonably well.

This week I started thinking about the general layout, design, and content of CeNCOOS's HAB web page.  I've been researching a lot of different educational sites such as NASA's Giovanni and trying to get a basic idea of how to present the information because I'm not an educator and I've never put together a web page before.  In addition, I contacted a few high school and middle school teachers to get their initial feedback on the material and sites I've been looking at.  I also got input from a couple of HAB researchers and am hoping to use some of their data sets for the site.  My goal for this project is to provide improved access to all the HAB info out there so that teachers can easily develop interesting and integrated lesson plans as well as give students a chance to come up with some innovative science fair projects and reports.

Next week will be a continuation of emailing teachers, getting their feedback, mapping out the web page, meeting with Raphe (one of the top HAB researchers) and working with the computer guys to test this all out. It's still been difficult to get a hold of people, but as my mentors already pointed out, I have a lot of material to work with so I'm not too worried.  I have more than enough to chew on for now!

This week was a fresh start for the diphyes fluorescent protein expression, with a new vector and competent cell system. Meghan was also making new first strand DNA to use, as opposed to the old double strand DNA I had been using. The fresh 1st strand wasn't ready until Tuesday, however, so I spent Monday kayaking (thanks!) and aligning my sequences from the week before. There were at least six contigs (pair of forward and reverse sequences) containing my gene of interest, however, the protein purifications did not result in a fluorescent protein, meaning the E. coli were not expressing my fluorescent protein. This could mean that the sequnce was effectively trasformed into the vector and taken up, but the cells didn't fold it correctly so as to express it. Anyways, this week was spent working with the new vector system and chemically competent cells a new technique I'd never learned! Meghan is a really great teacher, very patient and uncerstanding. After lowering the annealing temperature for the new primers, the PCR finally worked! That was purified, digested, ligated, and transformed into the chemically competent cells (two different kinds) and i got colonies. i just induced them on friday, so I dont know if they're glowing yet. I grew up cells for protein preps and sequencing at the end of the week and have the pelleted cells to work with on Monday. Looking forward to seeing these cells glow! Also, camping this weeking was really fun and relaxing

Principal component analysis rejected by advisor.  Keep it simple.  Time to decide if this aspect of the data is worth pursuing.  Its coming up on sink or swim time.  I'm not entirely at ease.

I disagree with a lab group member about one analysis method.  I think it is not rigorus enough.  People have worked their entire carriers on a data set which I have looked at for five weeks.  I hope I am wrong, but understand that I cannot use a method that I disagree with.  Talking with advisor about this on Monday. 

I've shoved all the data into a 4-D space because its fun and convenient to work with and I haven't yet found any excuse to use any higher dimension (although I dreamed of one last night).

The Best Week:

            Successfully completed my first task of passing messages from one endpoint to another through Enterprise service bus. Eventually I was assigned the very next task, which is transforming messages in ESB. Since XML is the standard message format, it is quite obvious that there is need for transformers in the ESB to convert any message format to XML and XML to application specific format. Moreover, I feel terrible thinking of how much time is left, and I think this is the best time to start working on the paper.

I'm finished testing the software I wrote to remote control the boat in simulation. I spent this week converting it over to use the boat's sensors instead of my development board's. Also worked on integration: I needed a bunch of connectors made and some soldering done before I could attempt. For some reason RS232 control of the boat isn't working and that's my main problem right now. I have also had a hard time finding pins on the development board that aren't already used by peripherals that I can use for inputs or outputs to the stepper driver board.

I also learned this week to not schedule a boat launch until you're really sure you can do it. I scheduled and scrubbed the launch 3 separate times which means communicating to 5 different people a change in schedule. Not very fun...

 This week I plan to finish integration of the remote control with the boat. I need to dry run it and THEN I'll actually schedule a launch time. I have a couple of problems I'm running into that I listed above but I'm confident I can get them done this week.

Hi everyone!

Totally forgot about my update on Thursday, so here it is.

It's scary that it's already week 6....I have to get something finished soon! I hit a snag with my OMZ project, but hopefully that's been sorted out. This week will help decide whether a big revision of my initial idea for the piece needs to happen. As time is short I hope it doesn't! Am looking forward to this week, but definitely feeling the pressure of a lot to do and not much time left. Hopefully this week I should have....

 The T-shirt design completed and all the ordering/cash stuff that goes along with it finished.

I'd also like to have a revised version of my OMZ piece done to see how the new data works with it.

 I want to complete the second panel of the illustration series I'm doing for Alena/the aquarium

I want to get the composition finalised for my OMF project (this will depend on whether I receive the reference materials in time, so I'm crossing my fingers on that one.

Finally I need to get started on my end of internship paper. Aiming for a page done this week.

Bye for now

Laura

I continued to work on the oxygen sensor project. Read a lot about oxyen sensors, how they work and the best practices for using them. Completed the winkler titration on the water samples we collected durring the test cast. Chased down a few blips in the data. Graphed the data which revealed some interesting and unexpected bumps and lags. Met with someone with experience in evaluation of oxygen sensors(Jason Smith of ACT/MLML). Started powerpoint presentation.

Next week, Will complete data summary and graphs for another set of sensors which was evaluated in June. Will chase down some calibration information and review my results with others. May complete first draft of presentation.

Really enjoyed Kayaking and sailing this week.  Also met with my Mentor for the first time and he gave me good input on how to choose a thesis project. Enjoyed this weeks seminar by Lars Tomaneck which gave me some ideas for a thesis project.

Looking forward to scuba diving.

This week started with a day of acting as "fly on the wall" in a meeting of Monterey Bay Aquarium managers and MBARI staff who went over next year's proposals - as always, my exposure to such meetings is a great way for me to learn the ins and outs of the relationship between the two institutions.  There are great ideas for next year!   Our ever important intern meeting took place on Slough waters in kayaks - who doesn't enjoy that!? We had some encounters with sea nettles which were just gorgeous in the water and of course, plenty of marine mammals.  I am trucking along, making progress in the creation of the aquarium's new high school lab called "Carbonated Seas: an Investigation of Ocean Acidification".  I met with Dr. Jim Barry and handed him our program "packet" in the hopes that he will offer some valuable feedback - next week I meet with him for a longer period of time to get the scoop on ocean acidification research at MBARI (I hope I can find mini tapes for my recorder!).  My program script is almost complete - I found that it took me longer than I planned because there are a few areas of this topic that I am unsure of, so of course, I had research papers and text books strewn across George's desk (he was out!) which I had to stop and refer to.    I had a very exciting meeting with Todd Walsh who will work with me to create a movie for the introduction of the lab!  What an incredible resource he is!  I am spending Thursday and Friday at the aquarium with a fellow educator who is also working on the ocean acidification lab.  It's always good to see my colleagues  and of course my animals!