International Symposium on Learning 2006 - Day 2 - Part 2

Day 2 - International Symposium on Learning 2006 part II

Friday, December 01, 2006

10:08 AM

 

International Symposium on Learning 2006 sponsored by KAST

 

http://learning.kast.or.kr/

 

Yoshihiko Nakamura

University of Tokyo

 

What do Humanoids Learn from Humans?

 

IRT Foundation for Man and Aged Society (project) to research a broad range of areas involving robotics

 

Humanoid - human-like (resemblance) robots.

 

Discussed the modeling of humanoids by looking at the human anatomy. In particular, the framework of bones, joints, ligaments, and muscles.

 

This is interesting, but where is the learning science connection? When do we talk about learning? If I wanted to build a humanoid, I would be loving this, but this isn't really about what humanoids learn from humans, it's about what robot designers learn from the human body.

 

 

 

 

International Symposium on Learning 2006 - Day 2 - Part 1

Day 2 - International Symposium on Learning 2006

Thursday, November 30, 2006

4:38 PM

 

International Symposium on Learning 2006 sponsored by KAST

 

http://learning.kast.or.kr/

 

Raja Chatila - Learning Robots: From spatial cognition to skill acquisition

LAAS-CNRS

Toulouse, France

Raja.Chatila@laas.fr

 

What is a cognitive robot?

o        Integration of perception, decision, and action

o        Learning concepts and interpreting the environment

o        Deliberation and decision-making

o        Learning new skills

o        Communication, interaction, and language

 

Robot companion - European Project COGNIRON

http://www.cogniron.org

 

Learning Requirements:

o        Objects

·         Multi-sensory, 3D, object modeling and recognition; from view-based to object based

o        Space

·         Maps, regions, concepts. Appearance, geometrical, topological labled models, landmarks

o        Situations

·         Spatial and temporal relationships

 

Spatial mapping requires a combination of object and topographical processing.

This involves incremental mapping that the robot learns over time.

 

Beyond spatial toward communication. Really talking about a sort of communicative competence. Takes signals from the environment and interprets them to devise appropriate responses.

 

Object modeling. First you need to recognize items in the environment. This requires constant processing of environmental data, including the use of 2D tracking and 3D representations.

 

A lot of training is required. This is similar to training voice recognition or even handwriting recognition. They started it with videos of people doing a series of actions. This is then interpreted by the robot via a 3D representation of the human.

 

Move to autonomous learning.

Learning concepts to learning skills

o        Open-ended

o        Common representations

o        Process guided by utility

o        Incremental learning

 

Interesting building of temporal knowledge. The robot stores information "maps" about an object from multiple perceptual angles. These maps are then combined to enable the robot to recognize the object at any angle.

 

Multiple object recognitions can be combined to recognize groups (scenes) of objects. This is similar to chunking in language learning. Learning to group items for easier production, or in this case recognition.

 

Provided a cognitive chart at the end, which would have been an hour discussion in its own right. I wish that he could have spent more time on it giving this audience.

 

Take away - Learning about ones environment is really a precursor to interacting with humans in any sort of naturalist way. To an extent, this is entirely possible at this point, but will require a lot of work. Also, autonomy is still a ways off, but it's as much of a question of time/information as it is about technology. The building of communal knowledge.

 

International Symposium on Learning

 

Day 1

Thursday, Nov 30, 2006, 11:14 AM

Hotel Grand International - Seoul, South Korea

Here are some notes that I took on my tablet using MS ONENOTE.  That might explain that terrible appearance because these are all based on MS handwriting recognition.

International Symposium on Learning 2006 sponsored by KAST

1st speaker = Daeyeol Lee

Neural mechanism of reinforcement learning and decision making

* Monkey video

-sound of brain activity with different movements/decisions

* Matching pennies-studied via game theory (ie, Nash in beautiful mind)

-No behavior is random

-Animal must randomize choices, otherwise computer will play off of strategy. This is the reinforcement model

* Dors-lateral prefrontal cortex-association with learning behaviors

IX Essentially, it seems that fewer neurons fire after repeated trials

* Sondheim-seems a little like and" no duh" town Supp J conclusion

This could indicate that once connections are made, future decisions

require less of a load on the brain.

Conclusion – Seems like a “no duh” conclusion.

Overall, a great speaker. He really made the topic interesting and engaged the audience. Interesting research.

Amy Poremba -Learning and Memory in the Auditory System

Testing auditory signals and learning behaviors In rabbits.

* what is the brain doing during learning

-removed brain segments to isolate areas necessary for learning

if tone is accompanied by shock received in animal's foot.

* Sensory modalities are used at the same time (auditory/verbal)

* environmental attributes can effect learning. I wish that this would

have been clarified better throughout the presentation.

panel discussion

Operant Conditioning (evident in previous speakers work)

-law of effect - the response is a function of its consequences.

-theory of mind. ability to build models and guess whet others are dtp thinking.

this into is the used to predict actions and counter-measures

can be taken.

-Moves from Operant Conditioning to more of a Cognitivist theory of a processing model.

-Operant Conditioning cannot explain the thought process in game theory

Would it make a difference if the experiments were done with 2 monkeys as opposed to I monkey Vs./ computer?

-Yes, but this will take awhile to do.

Can your research explain decision making? (improve. decisions-making)

Is there a universal learning process?

There are many Similarities in animal models.

She essentially stated the operant conditioning has a major place in learning.

A Quick 30 Writing Tips for the Start of an Academic Career

Thanks to Curt Bonk for this tips. If there's anyone out there that knows what's up, he does.

Podcasting through del.icio.us

Sometimes you have to get beat over the head to get the idea. Today, I had one of these moments.

People often ask me about podcasting and I've continually given advice that puts them through hoops to get a podcast up and going. There are services out there that make it rather easy to podcast, but all services have their limitations, whether it be content, storage, cost, advertising, and so forth.

Most of these people want to podcast for their classes and they aren't concerned with a snappy interface or "face" of the project. They are very practical. They want to get the audio out there to their students.

One podcast that I listen to occasionally is an edtech podcast (I'm not even sure of the name), but you can access their files from their del.icio.us account (http://del.icio.us/edtechlive/mp3). I've been looking at that for months but I never put it all together. All I have to do is tag links to MP3s in my del.icio.us account in a way that groups them logically, then add the feed to my iTunes....tah dah! Podcast.

Now for the cool part. They don't have to be my files. They can be any MP3 file out there that I think is interesting. I can provide (push) this content to my students without them going to a website (pull) to retreive it.

I can do this from my blog by simply providing the del.icio.us feed and therefore connect my postings to the podcast. I can bypass the blog altogether and simple fill out the comments section in del.icio.us to describe each file.

I wish that someone had told me this before. Though, they might have and I just wasn't ready to understand it until now. This, however, is going to change the way that I talk about podcasting with my students and anyone else who will listen.

To start this yourself, sign up for a del.icio.us account now at http://del.icio.us

KTF's Video Calling Service Takes Off

Dare I dream? Will I finally get a video phone? Oh, so close.

I've thought that this would be possible for years, but I don't think that phone companies saw a benefit for them. They might have noticed that people want to do everything on their phones and a video calling service is only logical.

Japan (at least in Tokyo) has had this for years. I remember reading about it with a significant amount of jealousy. I also know that the US carriers are going to do this any time soon unless Skype succeeds to open up the wireless carriers to 3rd party hardware/software developers.

Ah, but now I'm just rambling. I'll sit back a minute and dream about my first video call.......

Teens Can Multitask, But What Are Costs? - washingtonpost.com

I think that this is potentially a major issue that we will have to consider when using more open systems in education. For example, how to I ensure that my students are "on task" in Second Life when they could be doing 10 other things online.

This is true of classroom-based and distance courses. If I'm online discussing topics with students I want them to "listen" to what I and other students are saying. If they are off on other channels, they are simply going to miss it. There is no true "multitasking". Something is always hurting. I, for one, have always studied with the TV on. It's a terrible habit, but I've been doing with since I started school. I have trouble studying without it. However, I KNOW that I'm not studying as well (as efficiently) with it on.

So, can you multitask? Sure. Is it as effective as focusing on one task and completing it before you do another? Likely not.

Like the researcher in the article says, it's all about depth. Shallow activities (checking out Web pages, listening/wathcing TV, checking email/blogs/rss, and so forth) can be done more efficiently in many cases. Web pages take time to load and other things take time to process. During that time to you can flip through another email.

Deeper activities, like reading a academic article require much more attention and depth of processing. You are not just gathering bits of information in these activities. You are gather information, comparing it to what you already know, judging whether it's worth keeping or not, deciding how it applies to the rest of your learning, filing away relevant aspects for future use.

If you can do that effectively and efficienty while managing chats, TV, and so forth you are amazing....and really too good to be true.