Think Again Networks and the Delta Loss
arXiv:1904.11816v22 citations
Originality Synthesis-oriented
AI Analysis
This work proposes a new abstraction for state-dependent functions, but appears incremental as it builds on existing recurrent neural network concepts without clear application or impact.
The paper introduces Think Again Networks (ThinkNet), an abstraction for state-dependent functions like recurrent neural networks, but does not specify a concrete problem tackled or results achieved.
This short paper introduces an abstraction called Think Again Networks (ThinkNet) which can be applied to any state-dependent function (such as a recurrent neural network).