boostmetric vs lmnn

gistfile1.txt

Initial -> Initial Accuracy 
Final -> final accuracy after boostmetric learning
Timings -> boostmetric timings
Final_LMNN -> final accuracy after lmnn learning
Timings_LMNN -> lmnn timings

Iris:
k   Initial  Final    Timings   Final_LMNN   Timings_LMNN   
3   96.000   96.000  0.063613   96.000        0.347348     
5   96.667   96.667  0.116031   96.667        0.501501     
10  96.000   98.000  0.308844   97.334        1.088812     

vc2:
k   Initial  Final    Timings   Final_LMNN   Timings_LMNN   
3   75.858   75.848  1.055748   76.811        1.245180     
5   77.778   77.778  3.357797   77.778        2.046515     
10  82.126   82.126  0.502626   81.642        4.962297

balance_scale:
k   Initial  Final    Timings   Final_LMNN   Timings_LMNN   
3   77.92    85.60   0.398939   91.20         4.193676     
5   80.64    91.84   0.663194   95.52         18.706549     
10  89.92    92.64   2.322994   92.800        20.26838

ecoli:
k   Initial  Final    Timings   Final_LMNN   Timings_LMNN   
2   82.14    83.33   0.230177   83.333        2.780234     
4   86.01    85.41   0.563959   86.60         6.035063  

covertype-5k:(for LMNN max iterations were set to 5 & for boostmetric max_iterations was 500)
k   Initial  Final     Timings       Final_LMNN   Timings_LMNN 
1   86.84    86.84     67.657751     88.16        16.054354     
2   86.84    86.84     267.37731     88.76        18.996322     

letters:(for LMNN max iterations were set to 5 & for boostmetric max_iterations was 500)
k   Initial  Final     Timings       Final_LMNN   Timings_LMNN 
1   99.595   99.92     84.91842      100.0        279.768357     
2   99.585   99.75     176.6085      100.0        296.327887