This commit is contained in:
Alan-Lee123
2017-06-16 01:14:22 +08:00
+2 -2
View File
@@ -1,7 +1,7 @@
# relation-network
implement relation network introduced by deepmind using keras.
keras implementation of [A simple neural network module for relational reasoning](https://arxiv.org/pdf/1706.01427.pdf)
Relation network is a noval neural network introduced by deepmind in [A simple neural network module for relational reasoning](https://arxiv.org/pdf/1706.01427.pdf). It can achieve super-human performance in challenging visual question answering datasets such as CLEVR.
I implement Relation network using keras and train it on a challenging visual question answering dataset called [Cornell NLVR](https://github.com/cornell-lic/nlvr). The training is in progress. The temporal test accuracy is 76.87%, which is much higher than the previous state of the art (61.99%).
I implement Relation network using keras and train it on a challenging visual question answering dataset called [Cornell NLVR](https://github.com/cornell-lic/nlvr). The training is in progress. The temporal test accuracy is 89.10%, which is much higher than the previous state of the art (61.99%).