On attribution of recurrent neural network predictions via additive decomposition

Keyphrases

• Recurrent Neural Network 100%
• Neural Network Prediction 100%
• Attribution Methods 100%
• Additive Decomposition 100%
• RNN Model 66%
• Decision-making Process 33%
• Vulnerability 33%
• Challenging Problems 33%
• Failure Cause 33%
• Text Mining 33%
• Generalization Ability 33%
• Local Approximation 33%
• Black Box 33%
• Input Text 33%
• Varying Length 33%
• Linguistic Knowledge 33%
• Debugging Tools 33%
• Text Segmentation 33%
• Contribution Index 33%
• Phrase Level 33%
• State-of-the-art Performance 33%
• Interpretive Methods 33%
• RNN Architecture 33%
• Flexible Interpretation 33%
• LSTM-GRU 33%
• Attribution Scores 33%

Computer Science

• Recurrent Neural Network 100%
• Interpretability 25%
• Approximation (Algorithm) 8%
• Experimental Result 8%
• Text Mining 8%
• Generalization Ability 8%
• Art Performance 8%
• Long Short-Term Memory Network 8%
• Debugging Tool 8%
• Text Segment 8%
• Linguistics 8%
• Decision Making 8%

Chemical Engineering

• Recurrent Neural Network 100%
• Long Short-Term Memory Network 100%