TUM-Thesis/content/outlook.typ
Marius Drechsler 2be84b715f Ich habe fertig
2024-08-21 21:23:33 +02:00

12 lines
790 B
XML

= Outlook
Upon the findings of this work, further topics might be investigated in the future.
Generally, the performances of both helper-data algorithms might be tested on larger datasets.
Specifically concerning the BACH method, instead of using only $plus.minus 1$ as weights for the linear combinations, fractional weights could be used instead as they could provide more flexibility for the outcome of the linear combinations.
In the same sense, a more efficient method to find the optimal linear combination might exist.
During the iterative process of the center point approximation in BACH, a way may be found to increase the distance between all optimal points $bold(cal(o))$ to achieve a better separation for the results of the linear combinations in every quantizer bin.