Communicating our Science

Communicating our Science
The Other Haiku (Nerdish)
The Other Haiku (Nerdish)
The Other Haiku (Nerdish)

For the Fulbright Science and Technology Award seminar that happened over the last weekend I made a poetic entry for the contest "Science for all".

Here is the Powerpoint with the Haiku - What is Reinforcement Learning: A poetic Approach

Below is the explanation of the presentation. The presentation does not  in any way cover all that is reinforcement learning. Just a taste.

What is Reinforcement Learning?

Reinforcement Learning is a subset of Machine Learning or Artificial intelligence in which, machines learn by performing actions in an environment and then receive an associated reward. Learning, in most cases, aims to develop a strategy (action choices) that will lead to high rewards/returns. Reinforcement learning is being used in multiple applications including, but not limited to:

· Looking for treatment options for HIV by scheduling drug doses in the correct quantities and sequence.

· Assisting in debt collection in New York City by optimizing when to collect certain taxes given that people are in arrears.

· Used in the learning mechanism the folding robot uses. Yes robots finally fold your laundry (a very hard task that humans do easily).

· Reinforcement learning is used also to solve games. Some of the best reinforcement learning game agents play better than humans.

· Using game theory reinforcement learning is also used in predicting/creating economic models.

Reinforcement learning is also interesting, in that it aims to solve the fundamental problem that is Artificial Intelligence. Having an agent learn how to act rationally in an environment. My specific research is concerned with trying to use multiple Reinforcement Learning agents to improve performance in tasks. This is termed Ensemble Reinforcement Learning, a more heads are better than one approach. For the contest I tried my hand at Haiku, with accompanying pictures,to communicate my science. Machine learning has also been used to generate Haiku.

Download: What is Reinforcement Learning: A poetic Approach

Something extra

Like that last chip in the bag. I got a chance to meet Steve Wozniak the day after we finished the Fulbright conference. I had named him as one of my top scientists (or engineer as he refers to himself) I admire. So here are the pics. He came to Rutgers to talk about "Disruptive Innovation"

Vukosi and Steve Wozniak @Rutgers (14 November 2011)
Vukosi and Steve Wozniak @Rutgers (14 November 2011)

He also signed my iPhone and HP TouchPad

Woz signed all of these and he has a stainless steel business card
Woz signed all of these and he has a stainless steel business card

One of the points I took from him about bridging the gap between consumption and innovation are:

[blackbirdpie url="!/vukosi/status/136134082009632769"]

The Video of the visit. I am shown asking a question for like half a second 🙂

Leave a Reply

Your email address will not be published. Required fields are marked *


This site uses Akismet to reduce spam. Learn how your comment data is processed.