Creed(o) of Procrastination

Ok. Look. I procrastinate. I procrastinate a lot. I have a reasonably sized talk tomorrow at Startup Edmonton (you should come! Details found here!), of which I have yet to finish, let alone practice. So instead of doing that I’ve decided to come and enjoy the ritual of an espresso at Credo Coffee. I’m not proud of my choices, but they are the choices that get made (either by me or at the whims of some biological process). So why not talk about the espresso, get it out of my system, then head back to work.

The Espresso

The first thing I noticed about this espresso was the crema. It was a nice, maybe not as much as I typically would like, but it was a good amount and persistent. After stirring, which you should always do with a nice espresso, I always take the… the remnants… Wait. You don’t stir your espresso do you…

Ok, look. You can say this diatribe only exists as a mad attempt to procrastinate further (maybe you’re right), but I guarantee your next experience with a nice espresso will only be enhanced by this small tradition. If you already stir your espresso, congratulations! You can move onto the next part where I pick up the string of thought rudely interrupted by those of you who don’t.

You should stir your espresso before consumption. If not for the ritual, for the taste. Long story short, when you initially get your espresso it is NOT homogeneous. There are three parts that make up the espresso: the crema, the oils, and the solids. I won’t go into detail in this post, but you can find a pretty good explanation here. By stirring not only are you lowering the temperature (which makes for a better connection with your palate), you are bringing together the three aforementioned parts to express the full potential of the brew. STIR YOUR ESPRESSO… Or at least stir good espresso…

Back to it. I always take the remnants of the stir on the spoon and have an initial taste. Here the crema hits hard. It is bright, and slightly acidic in the best ways. I dig into the body, and find a full bodied more savory taste. It matches well with the acidic overtones. It is a solid experience, and well worth the trip.

The Cafe

I like Credo Coffee on 104 street. It is a decent downtown cafe just off Jasper Ave. The ceilings are high and the atmosphere is comfortable. If I’m here I often like to sit at the window hightables. This gives me the option to sit in their (actually quite uncomfortable) high chairs, or stand (which I often do). They also have ample photographs (which may or may not be local, hard to tell) scattered on the walls, adding to the decor. One of my favorite parts of the cafe is the high ceiling in the main sitting area (as you can tell because this is my second time mentioning it). For some reason I always feel more comfortable when the ceiling is high, probably just some weird quirk of my lived experience. Overall it is a really nice place to have an espresso and procrastinate… Erm… I mean… relax… in the middle of the day…

A bite from my Startup Edmonton Talk

As promised, I may sprinkle bits and pieces of AI into my coffee talk (and bits of coffee into my AI talk). Tomorrow I’m giving a talk on Reinforcement Learning, specifically the application of general value functions, to the AI meetup group. I’m planning on turning this talk into a full blog post in the future, but here is a bit to wet your appetite.

What is a general value function (GVF)?

A GVF is just a value function (this will be explained more fully in the future), where we are learning about observations of the world rather than only the reward signal. The main take away from this is that we can make predictions about the future stream of data given our current knowledge and history. Unlike supervised learning, GVFs are heavily grounded in a stream of experience and can be learned online without any labels! This can make for some really powerful applications. Probably the most canonical example is making predictions about the various sensors of a robot. The robot can ask “what is the likelihood of me maxing out my motor?” With this question answered as a prediction the robot can make better decisions about applying more voltage to the motor currently, so to protect the future reliability of the motor.

Wrap-up

Ok. I’m off to work again. But if there is anything to take away from this post it’s this: STIR YOUR ESPRESSO!!! (also come to my talk I guess…)