[SPL] Comparisions with SPEA2 and NSGA2

[vivekaxl]

---

[vivekaxl]

Comments

1. The hypervolume and spread is improving (as if there is memory)
2. Variance in GALEX
3. Loss of GALEX is faster than others

In Cell Phone: Cost doesn't decrease for GALE X

[vivekaxl]

Possible Options:

Option 1:
[ ] Collect all the non dominated leaves over all generations -> Keeps the memory + effective sample spaces + has randomness of GALE (which we think is the most important in GALE
[ ] Do non-dominated sorting - use the top 100 points and then compare to algorithms

---