Minutes

During the thesis I will have several meeting. The minutes of those meeting are posted/updated here.

First Meeting

Opening: 30.11.2011 14:00
First meeting with M. Kaisers and K. Driessens.
We talked about the paper of TLS and they gave a little overview of has been done so far.
rough thesis appointments:
– Development of a TLS agent which is useful for testing purpose (Java is possible)
– Implement a second agent which is able to reuse knowledge during search
– benchmark both on benchmarks from literature
– benchmark both for the “circle-game”
– select a practical test and benchmark both

Appointments for the next meeting:
I have to read the unpublished Hoot paper and start with a thesis plan

Second Meeting

Opening: 7.12.2011 14:00
I did a first draft version of the thesis plan and read the Hoot paper.
I received the correction of both supervisors.

A few questions has to be clear for the next meeting:
– when is it possible to gratuate
– when does my elective semester officially end

Following dates were defined ( if they are possible )
Starting date: 1.2.2012
Hand in draft version: 29.7.2012
Hand in final version: 13.7.2012
Defence: 25.7.2012
(because otherwise we will run into problems with summer holidays and M. Kaisers journey to UK in September)

Appointments for the next meeting:
1. Start a blog about my progress of the thesis
2. Hand in he final version of the thesis plan

Third Meeting

Opening: 9.01.2012 14:00

Announcements

I had problems to read/interpret the Scala code of TLS
-> Kurt will handle my request via Mail and will talk to the author
I asked for more interesting/necessary literature
-> Except for one paper, the received papers are the core

Tasks and questions from last Meeting
– “when is it possible to graduate?” – always , there are no restrictions
– “when does my elective semester officially end?” – mid of February

Appointments from last meeting:
“1. Start a blog about my progress of the thesis” – Done
“2. Hand in he final version of the thesis plan” – Done (after this meeting)

Planning
There are four ways that could be tried for reusing knowledge:
– deleting
– cloning
– rebuilding old tree with legal actions -> there it would be necessary to save used samples

Possible benchmarks:
– 1d and 2d “one-shot tests” from the TLS paper
– “circle game” for the robot
– “Swing up” (keyword : aerobat ?) -> has to discrete 2 actions

Appointments for next meeting
– I have to implement/come up with a first software structure for TLS (ask Kurt)
-> Take a look at “RL-Glue”(testing environment). If it is only for discrete action space it is useless.
-> try to recap one “one shot” function (design question for testing environment)
– Search for my “standard” MCTS references from the second semester project
– keep on reading literature

Next Meeting/u>

26.01.2012 14:00

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s