Predictive Analysis of Business Processes Using Neural Networks with Attention Mechanism
ieeexplore.ieee.org/document/9065057
1 Users
0 Comments
2 Highlights
2 Notes
Tags
Top Highlights
As outlined in Section IV, case lengths vary greatly. For naive batching where every case is padded to the length of the longest one, this results in a large amount of unnecessary padding. Instead, we group cases into buckets according to their length and only batch together cases from the same buckets
To ensure similar processing times for each batch, we don't set a fixed batch size. Instead, we only define an upper bound for event per batch and vary batch sizes according to the trace length of the batch.
Ready to highlight and find good content?
Glasp is a social web highlighter that people can highlight and organize quotes and thoughts from the web, and access other like-minded people’s learning.