Posts Tagged ‘applied sciences’

Artificial Intelligence Q&A with Tim Winkler Follow-Up

photo (26) copy

The event was a big success with over 30 people showing up to hear Tim answer questions about IBM’s Watson. Here are some pictures and sample questions (may be heavily paraphrased since they are drawn from my shoddy notes and memory) from the event:

Tim at QA2

Q: Can Watson pass the Turing Test?

A: Watson has never been given the Turing Test.

Q: Are there plans to build a physical analogue for Watson? (Asked multiple times)

A: No.

Q: Can you tell us about the specific details of what you do?

A: No, but after Watson was on Jeopardy IBM released detailed documentation

Q: Has anyone fed Watson info about itself?

A: No.

Q: There seems to be a competition between IBM and Google in the realm of AI. Do you believe that the future will include more mainframe-based AI’s like Watson or decentralized neural network based AI?

A: I’m a big fan of decentralized neural networks.

Q: How do you go about getting a job in AI?

A: I have a CS degree with an AI concentration, and I got an internship with IBM that eventually led to working on Watson. I worked on unrelated projects before this. There’s no set path.

Q: Is there any project to work on improving Watson’s ability to interpret history?

A: There are many NLP (natural language processing) projects that focus on solving that problem.

Q: Does one version of Watson know what other versions of Watson know? (i.e. medical student Watson vs cognitive cooking Watson)

A: No.

Q: Why is Watson so much better than Siri?

A: Siri is not really an AI aside from its NLP abilities.

Q: What question do you wish people would ask about Watson?

A: You guys ask good questions.

Q: Do you do unit tests and end tests on Watson?

A: Yes.

Q: Are there any Easter eggs in Watson?

A: I can’t tell you.

Q: Do you have a button that stops Watson if it turns into HAL?

A: We’ve had no serious thoughts of Watson turning on people.

Q: Watson does not have ontological understanding of the world; any benefit to adding that?

A: We’re working on it.

Here are some related links to the Q&A that Tim shared afterwards:

1.)    “Here’s the IBM research journal issue on Watson, that gives away all the tech secrets anyone would want”: http://ieeexplore.ieee.org/xpl/tocresult.jsp?reload=true&isnumber=6177717

2.) Behind a paywall:( “Computational creativity for culinary recipes”: http://dl.acm.org/citation.cfm?doid=2559206.2574794

3.)    “This is relevant to cognitive cooking, Florian Pinel is one of the authors, he’s our team lead”: http://arxiv.org/abs/1311.1213

17

07 2014

Join Us For Some Science!

spring

Hands-On Science | Experimentation | Curiosity Knows No Bounds

Applied Sciences Chicago

This is a group for anyone interested in applied sciences, including optical and scanning electron microscopy, fermentation science, saponification, herbalism, astronomy and planetary science, and citizen science.

We started this Meetup because many people think science must be done in a lab with expensive equipment. We want to show how easy and fun scientific experiments can be. We will gather to do science, talk about news in the world of science, and have field trips to places like Yerkes Observatory in Wisconsin.

This group is open to people of all experience levels, from complete newbies to working scientists. Bring your curiosity and join us!

Also, please like our page on Facebook. Help us reach 100 likes by the end of July!

07

07 2014

Artificial Intelligence Q&A with Tim Winkler, IBM

Applied Sciences Chicago presents:

What will you do with Watson:

Come out this Monday evening from 7-8 to do a Q&A with Tim Winkler, a former PS:One member and software engineer at IBM who works on Watson, the Jeopardy winning artificial intelligence.

Tim has worked for IBM for 10 years and on Watson for about 3 years. He works with natural language parsing and machine learning.

Watson processes unformatted data, i.e. natural language documents, and not structured databases, so part of Tim’s job is to work on ingesting that data and making sense of it. One of his current projects is cognitive cooking, in which Watson comes up with recipes for us to cook and is really awesome:)

Cognitive Cooking in the IBM Cloud:

So come with any and all questions about one of the most famous robots in the world:)

When: Monday July 7, 2014 7-8 PM
Where: Lounge

More links to check out:

IBM’s Watson Supercomputer Destroys Humans in Jeopardy
Watson is headed for your pocket
Watson’s new job, IBM salesman
Watson goes to the hospital
Artificial intelligence

06

07 2014

Hack Your Coffee!

image_9

Have you ever felt dissatisfied?

Have you ever woken up in the morning, considered all the projects you had ahead for the day, all the worthless meetings and teleconferences, and just said, “blah?”

You take a sip of your morning coffee, expecting a moment of brightness, goodness, something to cling onto by the fingernails, to hold you throughout the day, as you say to yourself, “You know, today is gonna suck, but at least I have this freshly brewed Java Supremo from Overpriced Cafe”?

You take that sip… and you say to yourself, “Really?  That’s it?  This is how my day is going to begin, not with a bang, but with a whimper?”

Yes, I too have felt that way.  I had wished there was a way I could seek out bold new flavor profiles in my morning coffee, beyond just “mild” “medium” and “strong.”  I wished there was a way to hack my morning coffee.

Thankfully, there are options.  Today, we are trained to only look for coffee at our grocer, or perhaps at the local cafe where we can pick and choose from pre-selected and pre-roasted coffee, perhaps even pre-ground, for our approval.  Or, we can do things the way our great-great-grandparents did it, which is, we roast our own damn coffee.  There are many ways to do this, one of which is with an air-pop popcorn popper.

image_3

That’s right, this noble device, previously only a monotasked tool to deliver copious amounts of fluffed corn to your gaping maw whilst watching reruns of ‘Love Boat’ and ‘Magnum PI’ can also serve as a way to roast our precious coffee the same way the pioneers did.  Well, not really, but close enough.

Currently only one batch of green coffee beans has been successfully roasted and served to unsuspecting denizens of PS:1 (No, those are not flavor crystals, THAT’S REAL FLAVOR DANGIT) to glowing reviews.  This is only a proof-of-concept batch at this point, with much more work and experimentation to be conducted, but with continued testing and sampling, the overall goal is to create a viable procedure, optimize the process, and eventually create the capability at PS:1 to roast coffee, and tailor to your specific discerning tastes.

The output of this project will include a wiki page with optional training, an understanding of the variables of coffee roasting and how to tweak the process to serve your tastes (i.e. using APPLIED SCIENCE to understand the roasting process), and recommendations on where to source green coffee beans.

More information is forthcoming in a future 300 Seconds of Fame!

 

 

 

02

07 2014

Liquid Nitrogen – Part 1

LN2 withdrawal systemLiquid nitrogen (LN2) rocks. Yes, officially speaking, I need to get liquid nitrogen at PS:One so that we can do serious sciencey stuff, like, say, energy dispersive X-ray spectroscopy on the scanning electron microscope. But, unofficially? Driving nails into wood with a frozen banana, or making ice cream, is just plain awesome. And the clouds of chilled water vapor billowing out of flasks like something from a mad scientist’s lab… what can I say, I’m easily amused.

First, we need a source to buy LN2. Fortunately, a very generous individual, who happens to own a really cool company that has helped us in the past with CO2 tanks and refills, and provided free LN2 to test the EDX detector on the scanning electron microscope, offered us a great deal for purchasing it. But transporting -321F liquid to PS:One and storing it presents unique challenges.

LN2 boring the plugSo we need to start with a large storage dewar. These can be ridiculously expensive. I scoured eBay, found one, posted a crowdfunding request to the PS:One mailing list, and, thanks to the generosity of a bunch of donors, I made the purchase.

Read the rest of this entry →

26

06 2014

Lock Picking and SEM

spoolpinBack in October, Pumping Station: One hosted an event called Locktoberfest, an annual event run by the Chicago chapter of TOOOL (The Open Organisation Of Lockpickers.) It features, well, lockpicking, beer, and brats. (The beer is necessary to relax one’s hands. Really. I mean it.) A number of nationally known people came out to Chicago for this, including Deviant Ollam and Babak Javadi.

Lock picking involves manipulating small components. Small metal components, which are conductive, and would image well in a scanning electron microscope. And it just so happens that we’re probably the only hackerspace with a working SEM.

Read the rest of this entry →

15

12 2013

SEM, EDX and fun with liquid nitrogen

Our scanning electron microscope came with an Oxford Isis EDX detector that we were told was non-functional. After a little poking around, I discovered that the replacement power supply which supposedly didn’t work was shipped from London, where the default power is 240V. After changing the voltage, the computer suddenly recognized the electronics, and it passed all the self tests. That looked like a good sign, so the next step was to acquire liquid nitrogen, which is needed to cool the detector.

Fortunately, one of our members owns NFC, a company that, among other things, sells liquid nitrogen. He loaned us a dewar of LN2 so we could test it out. After transporting it back to the space, I asked Everett to watch from a safe distance and let me know if anything was spilling while I filled the dewar attached to the SEM. He took some video of the process. The plastic funnel I used was cracking as I was pouring, which in hindsight wasn’t that great of an idea, so maybe we need to find another solution here….

YouTube Preview Image

The detector took over an hour to cool down, but ultimately it worked beautifully! I kicked up the energy of the electron beam to 20 keV which excited the atoms in the sample to give off characteristic X-rays. The EDX unit measured the energy spectrum of the X-rays given off, and was able to suggest possible elements that have those peaks, which I could then label. The next day Susan Young, the microscopist who used this SEM when it was at its former home, came to the space to give me some advice on the EDX and the sputter coater.

At center is an aluminum sample stub, with a square of copper tape and a strip of carbon tape. The SEM is imaging an area showing all three surfaces.

At center is an aluminum sample stub, with a square of copper tape and a strip of carbon tape. The SEM is imaging an area showing all three surfaces.

After calibrating the detector on a copper target, I then tried imaging a sample that consists of an aluminum sample stub, copper foil, and carbon tape, that has some of each of these exposed. I’ve labeled three peaks for copper, one for aluminum, one for carbon, and one for oxygen. The peak at 0 is just an artifact of the detector. Here is a movie of the X-ray peaks building as the detector collects data:

YouTube Preview Image

Here is the complete spectrum:

EDXSpectrum

The EDX detector has the ability to determine not just what is in a sample, but where it occurs in the sample. I did this by defining energy windows, above. One for carbon, one for one of the copper peaks, and one for aluminum. Each time the EDX detects an X-ray whose energy falls within one of the bands, the EDX sends a pulse on one of several channels to the SEM. The SEM operates in X-ray mapping mode and, because it knows the beam’s position when the pulse is received, it makes a dot on a color coded map showing where that element occurs. This map is an overlay on the secondary electron image of the sample.

EDXMap

The aluminum peak is colored cyan, which dominates the upper left part of the sample. Magenta corresponds to the copper peak, which appears primarily on the lower left. Orange represents carbon. The detector didn’t detect that much of the carbon peak (seeing as it’s the smallest of the three), but orange dots are clearly visible on the right hand side. The surface in the middle is the edge of the copper tape, but it is almost vertical relative to the electron beam, so it doesn’t seem to be giving off many X-rays.

All in all, this is seriously cool technology.

21

07 2013

Pictures from the Scanning Electron Microscope

A few nights back, Brian and I took some images from the SEM. We exported them into TIF format, and then copied them via Sneakernet, a.k.a. using 3.5″ floppy disks and a portable USB floppy reader. I converted them into .png files. Click them for full 1024×768 resolution, the limit of the Leica image capture board. I’m very happy with how they turned out.

Pin Top

Read the rest of this entry →

09

04 2013

Scanning Electron Microscope Update

SEM and RyanBack in January, we got word that Philip Strong, a past member of PS:One, worked for a company that needed to get rid of a working scanning electron microscope and was considering donating it to PS:One. While we have an existing SEM in the space (a Leica S440, owned by JP, a member), this one supposedly was fully functional, had documentation, and we could get some help from the microscopist, Susan Young, who used it. Of course we were interested!

On Monday the 18th, I learn that yes, the donation was approved, but with a catch: It had to be moved on Saturday the 23rd! Read the rest of this entry →

27

03 2013