This is my utility post. If I have accidentlly published it, kindly look the other way–this is just a scrap of paper lying in the information supergutter.
http://www.xanadu.com.au/ted/zigzag/xybrap.html This guy has a good thing going on, and DESPERATELY needs a re-write of his mewling page. I will do it, gratis.
UPDATE: On second thought, his stuff doesn’t look open source. Puck ‘im?
Good to see you again, Dr. Kanamori! *** A Weak Learning Algorithm (WLA) is a simple one which gives a hypothesis with a high error value. Multiple WLA’s can be integrated to reduce the hypothetical error below what is possible with any single WLA. Some WLA’s can be integrated quite well, some not very well, and others not at all–that is, they do not work and play well with others. Boosting refers to the effort to boost the accuracy of the hypothesis given by a combination of Weak Learning Algorithms. The topic of this presentation is a method to determine the integrability of Weak Learning Algorithms in boosting.
Directions to PCG.
From the West exit of JR Shinjuku Stn: Starting at street level, in front of Keio Department store (the side facing Yodobashi Camera). Directly along the street beside Keio Department Store, you’ll see a number of bus stops. Go to bus stop number 20 and take either bus number 41 or 45 (200 yen fare). Get off at the third bus stop (Juniso-Ike-no-ue)- it takes about 6-8 minutes from the station. After getting off the bus, about 2 meters ahead, you’ll see a green pedestrian overpass. Go up the stairs and come down on the other side of the street, (walking in the same direction as the bus was going) within a few meters you’ll see a narrow street that goes off to your right. Walk down that street until you come to another small side street on your right. Go up that small street. On the left hand side you’ll see a 5-story building. That’s us, we are on the third floor.
Our address is: Pan Nations Consulting Group Co., Ltd.
Nishi-Shinjuku Forest 3F, 4-32-12 Nishi-Shinjuku,
Shinjuku-ku, Tokyo 106-0023
Tel: 03-5350-5875 Fax:03-5350-5876