danielbigham.ca: ai

Good Job Ray

10 Feb 2010 00:00:00

Ray Kurzweil has some pretty elaborate and seemingly unlikely theories about the future, but reading the following was a wonderful thing: "I get very excited about discussions about the true nature of consciousness, because I�ve been thinking about this issue for literally 50 years, going back to junior high school. And it�s a very difficult subject. When some article purports to present the neurological basis of consciousness... I read it. And the articles usually start out, �Well, we think that consciousness is caused by...� You know, fill in the blank. And then it goes on with a big extensive examination of that phenomenon. And at the end of the article, I inevitably find myself thinking... where is the link to consciousness? Where is any justification for believing that this phenomenon should cause consciousness? Why would it cause consciousness?" It's very rare that I read someone's thoughts about consciousness and think "Wow, this person really gets it, they're thinking about it in a very similar way to how I think about it". But I have to say, this was very encouraging.

Defining intelligence

16 Dec 2009 00:00:00

Thought of the day: Intelligence is the ability to form high level concepts from low level observations, and then to be able to relate those high level concepts to new low level observations to solve problems.

The prez made a funny

24 Nov 2009 00:00:00

http://gizmodo.com/5411752/obama-to-robots-im-watching-you lol, I love it.

Wowza

19 Oct 2009 00:00:00

Impressive? Very. Scary? Yes. http://www.wired.com/magazine/2009/10/mf_optigenetics

Starting at Wolfram Research

07 Sep 2009 00:00:00

Here I am in Boston, on Labour Day, and tomorrow I start working for Wolfram Research. When I first read about Wolfram Alpha last spring, I never imagined that something like this would have worked out. What a neat opportunity! Speaking about opportunity, my team has it's work cut out for it. As I've read what the press has had to say about W|A this summer, one thing is clear: The part of the system that most limits it's ability to answer people's questions is it's ability to understand user input. Much of the time, the system has the data and algorithms it needs to answer a question, but is unable to map a question to those data and algorithms. And of course, this isn't surprising, since it is really the language parsing layer of W|A that is the most difficult. Ever since computers were invented, people have dreamed about creating a computer that could understand questions -- and after several decades of trying, people become disillusioned with this vision. As of 2009, there are still decades of work to be done. It is a very difficult problem! In a recent report written by Stephen Wolfram, it is stated that the team has been able to reduce the "fall through" rate by 10%, which is fantastic progress for only 1 summer of work! Hopefully this trend continues, and I look forward to helping make that happen. Today we did a "duck tour" of Boston, which involved an amphibious vehicle... first driving around, and then boating down the Charles river. I even got to drive the boat, and let it be known that I was the only person out of about a dozen that could keep the boat going straight! What a hero. We also took a tour of Fenway Park. One of the tour guides said that we were the first group in about six months to be allowed onto the field. And when I say "field", I mean the warning track. They wouldn't let us on the grass. Think he was telling the truth, or just trying to make us feel special?! Eli has been fantastic for the most part, but he was pretty upset when we made it back to the hotel this afternoon... apparently, disrupting a baby's nap schedule is a recipe for one very angry baby! Poor guy. It will be fun getting up tomorrow, navigating the subways, and then arriving to meet the folks of Wolfram Cambridge!

Having fun with food

30 Aug 2009 00:00:00

Over the last couple of months I've implemented some more features for my Grace project. These features are focused around food, which was the original dream for the project, so it has been nice to finally get around to that. At first, I focused on expanding my database of foods and adding their nutritional information to the program. Next, I added the ability to tell the computer what you have eaten and have it draw a nice graph showing you what essential nutrients you are missing for the current day. Using the nutrition tracking has been an interesting exercise, and has confirmed my suspicion that I consume upwards of 5000 mg of sodium each day, twice the recommended amount. It has also been very helpful when certain other members of the family want to keep track of what they're eating without always having to look up points values. With a good food database in place, I was able to move on to more interesting things. The first step was to use the food database to create a recipe database, and thus to be able to calculate the nutritional information for any of our recipes. The second step was to allow the program to keep an inventory of all of the foods in our kitchen by saying, for example, "add milk to the fridge", or "add frozen strawberries to the freezer". The immediate reason for keeping an inventory was that we have so many cans in the cupboard (and things in our freezer) that it becomes very frustrating if you want to figure out whether you have a certain ingredient -- you end up pulling half the cans out of the cupboard searching. Now, you can simply ask "do we have any refried beans?", and if you need to narrow down the location, you could ask "where are the refried beans?". Although the inventory is helpful, the real money is being able to answer the question "what can recipes can I make?". For example, we were on our way back from church today, and as is usually the case, both of us were famished and didn't have much energy to figure out what we could cobble together for lunch. We figured we'd just grab some food on our way home, but instead I typed "what recipes can we make?" into my BlackBerry and a list of several options came back, such as Rotini and Fresh Tomato, Basil, and Parmesan. That sounded pretty good, so we drove home instead and had a nice home-cooked meal.

Grocery shopping

21 Jul 2009 00:00:00

Today was the first day I had been in a grocery store since spending some time working on some nutrition software. To give some background, I have written a subroutine so that you can ask "recommend a food", and the software will consider what foods will maximize your nutrition graph for today -- that is, to minimize the amount of vitamins, minerals, and nutrients still required for the current day, given what you've already eaten. I have noticed that this subroutine often recommends foods such as: - GoLean cereal - Raisin Bran cereal - milk - orange juice - cranberry juice - bananas - pears - yogourt - chicken breast - ... And so, as I walked along I had it in my head to pick up some GoLean, Raisin Bran, OJ, bananas, pears, etc. In addition to picking these items up, I was more interested in comparing the nutrition labels on various kinds of bread, as well as orange juice. I chose the bread that minimized sodium while maximizing omega-3... and I picked the orange juice with 30% DV of calcium; I got the PC variety rather than the Tropicana variety since it was $2.99 VS $4.50. I real breakthrough was finding Miss Vickies unsalted chips. They're just potatoes and sunflower oil, and they taste great! I can't believe it: A potato chip with 0 sodium! I have also started entering costs into the system so that, while maximizing nutrition, it will also minimize cost. Fun stuff.

Nutrition graph

08 Jul 2009 00:00:00

Here's my nutrition graph for July 8 2009: This is the output of my Grace project.

Nutrition

30 Jun 2009 00:00:00

I've been working on some enhancements to my Grace project that allows for queries such as "how many calories are in 1 Tbsp peanut butter?", or "how many calories are in {recipe}"? (And likewise for other nutrients such as fat, saturated fat, trans fat, carbs, fibre, sugar, protein, vitamin A, vitamin C, etc.) It also allows you to say "I had 3/4 cup raisin bran", etc., and then later on to ask "what have I eaten today?", or "how many calories have I eaten today?". (And likewise for the other nutrients) As I've been collecting nutritional information for various foods, I'm reminded of how thankful I am for laws that have been passed to promote consistent labeling. (Thank you government!) One complication is that I'd like the actual number of g/mg/ug for the various vitamins and minerals, but the nutrition labels give a "% Daily Value". I did a Google search for the Government of Canada web page, but I was unsure whether these percentage values were based on the recent RDA/AI (recommended dietary allowance /adequate intake) numbers. I ended up sending an email to the link on the "Contact Us" page and within 7 minutes I got an email reply back with a bunch of helpful information and a link that answered my precise question. Wow! I'm impressed.