New objects/additions V 1.6

New objects implemented:

• dualperceptron  - Kernel Perceptron
• ord_reg_perceptron   - Ordinal Regression Perceptron (Shen et al.)
• splitting_perceptron - Splitting Perceptron (Shen et al.)
• budget_perceptron    - Aggressive Perceptron, i.e. online & sparse, on a budget (Crammer et al.)
• pmg_mds - Pre-image with multi-dimensional scaling
• pmg_rr - Pre-image by regression
• rsc_burges - Reduced set via gradient descent
• rsc_fp - Reduced set via fixed-point methods for RBF kernel
• rsc_mds - Reduced set via multi-dimensional scaling
• rsc_learn - Reduced Set via learning

Small additions / bug fixes / issues:

• changed myexist() function - loses some (rare) functionality but fixed a huge slowdown on some systems caused by searching to see if filenames exist
• fixed isdeferred()  problem - can now do e.g. train(toy) again 
• fixed bug in multi-class @fisher algorithm (by Kiyoung Yang) 
• r2w2_sel  - now works with RBF kernels (thanks to Chirag Patel) 
• added two new kernels: triangular and linear with bias