Closed Tsreenivasan closed 7 years ago
Hi All,
Would we be using the Laplacian k-smoothing for the initial probabilities as well?
Example: k=.001 test_tags = [ ['D', 'N', 'V'], ['D', 'N', 'V'], ['V', 'N'], ['D', 'N', 'N', 'V'] ] |V|={D,N,V}
Would the calculation for the start probability be:
P(Start=D)=> (3+.01)/(4+3*.01)
Can anyone confirm?
Regards Tejas
Start probability for D is correct
Thanks Ranjeet.
Hi All,
Would we be using the Laplacian k-smoothing for the initial probabilities as well?
Example: k=.001 test_tags = [ ['D', 'N', 'V'], ['D', 'N', 'V'], ['V', 'N'], ['D', 'N', 'N', 'V'] ] |V|={D,N,V}
Would the calculation for the start probability be:![image](https://cloud.githubusercontent.com/assets/11288412/23143986/30963800-f78a-11e6-9665-d8c6e508e732.png)
P(Start=D)=> (3+.01)/(4+3*.01)
Can anyone confirm?
Regards Tejas