[mlpack-git] (blog) master: New blog post (64511e6)

[gitdub at mlpack.org]

Repository : https://github.com/mlpack/blog
On branch  : master
Link       : https://github.com/mlpack/blog/compare/24f64d1453ea5e8e281d863fed9ab9b134061199...64511e6b569ec1a17bab35e0773690d7d97e6bcc

>---------------------------------------------------------------

commit 64511e6b569ec1a17bab35e0773690d7d97e6bcc
Author: Bang Liu <bang3 at ualberta.ca>
Date:   Sun Aug 21 17:29:03 2016 -0600

    New blog post


>---------------------------------------------------------------

64511e6b569ec1a17bab35e0773690d7d97e6bcc
 content/blog/BangWeekFourteen.md | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/content/blog/BangWeekFourteen.md b/content/blog/BangWeekFourteen.md
new file mode 100644
index 0000000..44f9c5a
--- /dev/null
+++ b/content/blog/BangWeekFourteen.md
@@ -0,0 +1,5 @@
+# Neuroevolution Algorithms Implementation - Week 14
+
+During the last week, we have changed the design of HyperNEAT to avoid class inheritance, and implemented it and its first test case: XOR. We are now debugging this algorithm as it hasn't pass the test. Until now, we have figured out multiple critical bugs through the debugging of XOR. One is the Genome classs's Activate() function. Previously it may skip some neuron's activation due to the disabled links. Another is NEAT's MutateAddNeuron(). Previously it will generate duplicated links. After we figured out why, the problem is now solved. We have re-run previous NEAT's test cases, and they passed.
+
+Now the CNE algorithm is ready to merge once we adjusted its style. The NEAT algorithm may be merged after debugging HyperNEAT. As we may figure out more potential bugs during the debuggingof HyperNEAT, and thus we can optimize it.
\ No newline at end of file